When in doubt ask your local rigger. He will tell you what he needs. If you ever give zremeshed models to your rigger he will appear in your bedroom at 2 AM wailing "Why, oh tell me WHY??"
If you were animating, the heads on the left and right would be no-go areas. Triangles and ngons deform the worst, and too many polygons will make weight painting harder, and make your scene run slower. Middle one gets stars
So, here's my 2 cents: I really wish companies would invest in AI trained retopologizer, as what is seemingly happening with rendering RnD today - AI trained cloud renderings or GPU-based denoisers to name a couple of examples. No offence to someone who likes it, but it's just a mundane thing that eats too much time as of now.
I think tools like this will come in the next 10 years, for sure. Currently most of the retopo in the big studios are done overseas in India and China, as you're right - it is a mundane thing to do and it eats a lot of time (from a studio scheduling point of view). We already have tools like Wrap3 which makes the job a lot easier when doing humans. /H
Have you seen this Siggraph vid ruclips.net/video/H8K5CyQB_kc/видео.html. Smart Retopology tools might look something like this, essentially automated retopo tools with a certain level of human input. This gives you the best of both worlds. Plus if you're feeling lazy and the asset doesn't require too much care you can just let the automated tool do the job.
I love you guys. Every time I look at your content I learn something, 90% of the time it's just something you say offhand. This time I learned about wrap3. Very cool
Good topology is underrated! I feel like the most underrated aspect being how useful good topology is for sculpting. I sculpt mostly for illustration and retopo manually every time because it's so useful to have those low subdivision levels flowing how you need them to.
good topology is the heart and soul of polygon modeling, which sculpting is 100 percent part of imo :) Totally agree about sculpting on good topology, everything JUST WORKS BETTER.
Artificialhype dude, I did watch the whole thing and the general view is the same as mine. I am in agreement with most of the video. You seem to think I made a video to criticize this one, I quite clearly said I "respectfully disagree" with some points. "Respectfully" because I respect the point of view given and we're essentially on the same side but have a different opinion on a few nuances, and I don't disrespect theirs. You're the only person arguing *for* bad topology here. And it's clear you're unwilling to look at this from any point of view than your own so this is where I end the discussion and wish you good luck in your topological ventures *tips fedora*
Thanks for your input, Danny! Big fan of your work :) You're spot on - clean topology can be amazingly useful when doing illustration too, as the lower levels allows you to really get a clean base. I find that when doing stylized work, that it's rather easy to get messy shapes in ZBrush, so a good, clean base can help here. /Henning
First off, great points to bring to light again in an era where technology has improved exponentially. Of course everyone is entitled to their opinion...from their experiences or what works efficiently for them. If there are "issues" with their retopologized sculpt, what may be a "difficult" fix for some might be an "easy" fix for others depending on their knowledge and needs. I've been in the game industry for over 17 years so I come from the low poly (box modeling first) era where "manual topology" was all you could do. But as with most artists, I have evolved my 3D character workflow. I create models for both games, illustration and sculpts for print nowadays. Before I jump ahead, I think it is very important to have an understanding of manually retopology of a hi-resolution model, and at times, even certain props, vehicles or environments. It obviously gives you more control to have efficient polygon count, deformation and a better flowing model overall...especially in game engines. But on other hand, I believe that Zbrush gives the artist a freedom...a freedom that they didn't have with other 3D modeling packages; to quickly prototype/conceptualize characters without having to worry initially about deformation, polycount, etc. Zbrush art history has shown that there can be great character sculpts even from pros who haven't manually retopologized (or maybe haven't gotten to it yet). It seems the bigger argument here is when to do manual retopology. Some prefer it earlier but I have preferred to do it much later towards the end of the pipeline. I give my arguments below. Auto-retopology was never meant to replace manual retopology. It was meant to get the artist some "almost instant" results while you are sculpting to better flowing topology than it was in its prior state. It would take you somewhere between 30 minutes to hours manually retopologizing a character (depending how complex it is). I found that using tricks in auto-retopologizy will get you nice deformations so can to continue sculpt without losing time or worrying about sever topology issues. Sure it's not perfect but it's very quick and workable enough where its not difficult to manipulate if you run into "topology" quirks. That's where auto-retopology tricks comes into play to help speed your workflow along. I think doing manually retopology too early can be counter productive at times because, as we know with clients, there will be many changes to a character...sometimes even drastic where you would have to do manual retopology again. So now you're spending another extra 30 minutes or hours manually retopologizing again. So there are several cases where manual topology treatment is best served until close to the end. Time could be spent better elsewhere down the pipeline. Topology is very important but the client is not going to scrutinize incorrect topology on a character concept or translation when you present it. All that matters in this stage with Zbrush is to quickly get the translation to the 3d model; the likeness or appeal of the character, color, flow, pose, silhouette,...maybe even textures, lighting, etc. As @FlippedNormals showed, you can barely see a difference and manual topology is not as vital when it doesn't matter to the client....or even to artist at this point. Perfect topology is not necessary in this part of preproduction but you still have to have better working topology than you had. That's where knowing how to get nice flowing topology quickly with zremesher will save you a bunch of time. We know that the results of auto-retopology are still unpredictable even with the current technology but it is a godsend when you know how to use it quickly for your advantage when you need it most. I think the stage of manual retopology for games, films, illustrators and digital sculptors comes down to their preference. I think it comes down to this. If the artists feel they have better control doing manual retopology earlier, then that's the workflow that works best for them. If they feel that manual retopology too early slows them down and like the quicker workflow of tricks using auto-retopology then that perfectly fine too. Now not retopologizing at all? Not my cup of tea but good luck. Also, another reason I think it is important to learn proper manual retopology and not just rely on zremesher (or other auto retopology methods). Learning manual retopology (like subdivision/box modeling) will give you a foundation of how the polygons interconnect for proper edge flow and deformation which is important for games and film. Sure it's nice to use the quick tools that will "get you by" but I think it's good as a 3D artist to have a foundation and better understanding of topology. This is not a great example but it would be like a character animator or character illustrator that hasn't learned the foundations/principles of animation or drawing. An understanding of proper manual retopology can keep polygon count efficient too since you have more control. Also, in regards to characters that require low topology, there are certain "configurations" (poles, ngons, closed edgeloop, linked edge loop, etc) that not only look better when viewed in game engines ( or even animated film). I remember the joys of turning edges in low polygon modeling to get a smoother looking topology even for static objects and characters. This applies to non-gaming and non-film digital sculpting too but not as strict as game engines. True it may be less of a worry to manual retopologize, but you can get some workable topology in Zbrush with some auto-retopology trickery. But like I said, although I opt for manual retopology in the early phase of production..... time constraints usually limit how much time I want to spend on one phase of the pipeline. In my case, I save the manual retopology for the end. We know the product or character will go through several iterations and even total redesigns that manually topology would be better saved until the very end when it is approved. It doesn't make any sense spending time manually retopologizing twice when you don't have time to spare. That time could be better spent in posing, texturing, pbr creation, rendering, etc. Nowadays I mainly use the mask/polygroups zremesher edge loop creation trick (I first saw it from a Steve James tutorial). It's not perfect but it's a time saver that I have adopted into my workflow when I don't have time to spare. Another side point is that I think posing the character as fast as you can in Zbrush is so essential because you can see some possible issues in your model you might not have seen in the T-Pose. Manually retopologizing several layers of meshs (apparel, hair, props, etc.) can get daunting and very time consuming with something that needs a quick turn around to the client. That's where knowledge of the tricks and tools in zbrush can be a time/life saver like t-pose master, polygroups to guide zremesher, and even reproject to dynamesh at times. And even if you end up having a noticeably bad topology in one area but zremesher doesn't help, then just retopologize that area. I personally wouldn't have any difficulty or loss of much time tweaking portions that didn't deform properly if I didn't do manual retopology at first. ( I'm mainly referring mainly to pre-production models where posing your model quickly for approval is necessary for some clients.) The minor adjustments that I had to tweak with the sculpted pose or modified character was light years faster than the time taken to manually retopologize the whole or even part of a character...not to mention the apparel. If the client requested a major change, redesign or other extreme modification, it would have have to retopologized again anyways, so time spent manual retopologizing would've been wasted. If done properly with zremesher, no client is going to be concerned about imperfect topology and edge loops. So like I said, I'm a big supporter of manual topology I feel it's one of the last phases of the character production pipeline. Thanks for listening to my rant. LOL
Hey! Thank you so much for your very interesting and enlightening comment - it's much appreciated. It's fantastic to hear the opinions of somebody who's been deep in the trenches in the game industry. My experience in the film industry is very similar to yours, in the way that good topology is really a basis for everything, as it makes life a lot easier. That said, it's not always doable from a time perspective, and then Zremeshing stuff can be a great time saver. Essentially, use whatever tool gets the job done, without religiously sticking to one technique or tool. Again, thanks for your elaborate insights! :) /Henning
Thanks guys. I didn't think I was giving any helpful insight and just rambling away. LOL This was a great topic you brought up though. It's getting some constructive and heated reactions. hehe. I'm sure many more qualified industry pros....unlike myself ( I still have lots to learn) will have much to say on this. I just found you guys within the past year and your videos and website rock! They are very helpful and insightful as well. Thanks again! :)
This right here. All that talk and they never actually showed an example of it in practice so someone like me who has close to zero idea what they're talking about could actually see it doing its thang.
As a game environment artist, the only time i do manual retopo is if i have very, very few assets to produce. Modern baking and material painting tools are so good it really doesn't matter if your uvs and topo are awful. For many environment assets retopo can take several times longer than the actual asset so its really, really not worth it. I either use custom mesh normals so the mesh holds up on its own without a highpoly or i use automatic retopology. Its a good skill to have but its also a massive production bottleneck for environment art. =/
The topology has a role in making human readable texture maps in regards to island layout and separation, which allows for quick touch-ups in 2D software. So there's that. (Again depending on what workflows are preferred.)
For environment pieces that makes total sense. Good topology is for deformations and shading, but with environment assets that don't deform and whose shading is based on a normal map or custom normals, the best method is whatever is fastest and lowest poly. Agreed 👍
Its useful, but rarely worth the time. Environment topology tends to be different than character topology, like triangles are totally ok. It saves polycount, makes vertex painting smoother, can remove intersecting geometry which is more expensive to render in games, etc... but none of it makes a big difference and can often take longer than actually making the asset itself.
OMG, so cool. Ive waited for such a discussion or video about this exact topic for so long! Thanks again guys. A joy to watch and hear ya. Channel has become really awesome.
In topology edges should follow stretch lines. No more information needed. Also I love very low poly topo for animation. It's such a nice feeling when in the end your mesh looks almost evactly the same when you apply subdivision as it looked before XD
Good topology is incredibly useful for many aspects of 3d production, but there is always exceptions. Most animators and even some Mograph artists would appreciate good topology when working with models that are rigged or being put through some form of animated motion. Topology defines the model's ability to bend, twist, and sometimes even interact with the surrounding objects in the scene properly. Good topology can also be quite useful when any topological customizations are needed in the workflow. For example, new textures or geometry additions on the fly are much easier to create if the topology is logical and neatly organized on the mesh. Sure we have specialized software now with which retopology or painting without defining UV's is possible, but not everyone has immediate access to these workflows and, in a pinch, it is just so much easier to work with clean geometry and edge flow than messy or unnecissarily dense polygons.
Great video !!!! I used to spend alot of time building really nice topo, until zremesher came out. For a long time I was of the opinion that topology was dead. That was until I went back and started reworking some of my old models. I found that having nice topology allowed me to stay in lower sub divisions for longer, (especially for areas like eyes/lips or anywhere with alot of folds happening) while still maintaining lots of control on my geometry. Of course if you are doing hard surface it doesnt matter as much. Now I kind of feel like there are times when zremesher will cut it, but for most of my workflows (character artist) I much prefer to have good topo from the beginning. Plus remeshing can be annoying to get the results you are happy with.
Good video guys! I thought your “center head mesh” had 5-point stars to close to the mouth and then you called in out near the end of the video so I’m glad you did that. I also think your nasal-fold edge is a little high above the nostril, but I know everyone has an opinion on topology, so I make mine based on where I worked and what I see some “top” modelers make. I also know it can change… I have worked at a company where all the topology had to be quads, and as you said it was because of their UV software. I think it’s good to learn to model in all quads, and then if you’re allowed to use triangles, then you put them in non-deformation areas. Another thing not mentioned is it good to have stars with more than 5-sided-edges and having a star more than 4 edges on the center-line…? Again, good video guys and this is a subject I could talk about all day... :-)
Thanks a lot for your input, Roger! I know which company you speak of, and they sure know what they are doing :) The base mesh we are using here could definitely use with some improvements, like you rightfully point out. It was pretty handy that it had some errors though, as it meant we could use it as an example of how to improve topo. Good seeing you on our channel, and I'd love your input on our other videos :) /Henning
Hi there! I am an environment artist and I would love to see a tutorial/lecture about how you guys organize/cleaning up a scene after/during modelling everything. (naming convention, material naming, file naming, etc.) Thanks a lot!
Nice discussion, thank guys! I've learned a lot by watching this, topology is really important and automated retopology will never replace manual clean and logically setup topology, I agree. Cheers guys!
Voxelized surfaces I find are the best for painting with photo-scanned materials with displacement. You need the geo non pinched edgeloops and even sized faces, otherwise you'll most likely get texture stretching or other artifacts.
@@Andreopimentel a hobbyist that has been working on 3d models (and having completed like 5 or 7 more models since this comment) i can assure you. remesh seems okay if its a static simple model. "okay" being the operative word. it doesn't seem to be that fantastic. i feel like manually retopologyzing is the way to go almost always. it even opens you to make a subdivision surface modifier and a shrinkwrap (at least in blender)
@@taylorfaith2320 yes, but the zmesher guide tool allows you to select where and how the new topology will be created, have you had any experience with that?
@@Andreopimentel afriad not . keep in mind. Zbrush is still a substantial purchase. and as i said . i'm a hobbyist you might find better answers from another person in these comments.
Good topology is important for a few reasons. Animation is the most obvious. But in gamedev, it also acts as an optimization technique. The first is a properly topologized model can use much less geometry while maintaining shape in animation. The other bit is polygon fans, which is a micro-optimization that allows the GPU to blaze threw geometry without wasting time in swapping states
Isnt poly count not as a big deal anymore though? obviously you cant just have a massively high poly count on any thing,. but within reason? like a full character with 90k tris isn't out of the realm of acceptable for modern engines for PC games. mainly talking bout characters here. in game assets like tables and buildings should be low poly as possible , and modular if possible.
thanks for sharing your thoughts and insight on this topic and i completely agree with everything you guys said:). As a fellow modeller (4-6 years so far of learning, and starting to get into more character anatomy/sculpting ) and now some experience in character rigging, i found that lately after I created a face and manually retopolgise it. I tend to start testing blend shapes/skin weights, even though i am not entirely great at animating, to show me if the retoplogising i have done actually works or not. But this ofc is excessive to do in a production with the amount of time it takes to set up. I just been doing it lately as planning on animating a personal project by myself oneday soon😅
Nice discussion here! If you're gonna animate/deform something, DO MANUAL TOPOLOGY! You want to control where the main loops placed. You want to control where the fold loops are gone be. For all of these and much more, please do manual topology!
Dead on... When I am doing animation and characters (which I am often doing) manual topology is KING!! The only time I ever use ZRemesher is for the ears, with carefully placed curves.
Amazing video! I'm a game designer learning blender and was struggling so much with trying to eliminate every tri or pole... this video was like a nice ice tea on a hot day. Thank you SO MUCH!!!
It's not only the rigging. There's one situation where a good topology is imperious, and that's when you want to build the uv map. With bad topology, you will have a hard time to create a proper uv map. But there's a workaround, at least in zbrush- you can remesh, sculpt and get your perfect high res model, and then project it onto the low poly one. All the details will go into the normal map.
I work on games, all the things that you guys say still holds true, except yes we can use triangles anywhere, there is no smooth so no artefact in deformation, and in the engine everything is a triangle anyway. On top of that, because the polycount is limited we sometimes need to define the triangulation of a quad to make sure the silouhette looks smooth. Cheers!
Good topology is needed for characters. For hard surface it depends. It depends on if you can optimize the asset by a substantial amount by doing it manually. If you feel like it won't matter, then just Zremesh. Reto is very helpful for texturing things and for controlling texal density.
I use triangles and n-gons a lot. Just subdivide once and in a place where you had a triangle you'll have a tri-edge vert, and where you had a pentagon you'll have a 5 edge vert etc. The main "danger" is concave n-gons. the center of subdivision for those a lot of the time ends up outside of its perimeter and that is the main problem. Quads are just way easier to control and that's why you generally want only quads in your mesh.
The use dictates the mesh via the technical direction, but the crafts person defines the skill everyone needs to work with what they made, and the artist defines the look that needs to stay established. Everyone needs to be a generalist in a way in that you need to be part TD, part crafts person, part artist if you want to work with the orchestra that is animation. Because animation is not one discipline but many under a title. Great discussion but a good follow up referencing this is to cover modern retopo tools, what works, and what doesn't.
Im fed up with these videos about topology when nobody explains how to do good topology. What is the methodology behind it and how can i topologise correctly? And how do you get manual topology to have such perfectly straight loops? If anyone has any resources please help me.
Yes it is. Of course for animation but even when modelling it just makes things easier which means things move faster. As for environment assets, to get those nice smooth highlights run across an object nice topology is a must.
Good typology is necessary for two reasons first for your uv texture maps etc as well as animation but it all depends on what the project is really but it's best to make it a habit to have good topology flow on whatever it is...simple as that...
Yes totally agree that good topology will depend of the type of project and what will the model will be used on, but i have an interesting question for you guys, and if you could make a part two with this idea that would be awesome. The question is about topology styles, like you mention video game topology is different obviously than animation, and i see a lot of people with different styles of retopology, what are your thoughts on that topic???? Awesome videos guys!! big fan of yours!!!
Great question! So there are two main things to tackle here: - Topology which will be simulated. This topology is very even and gridded, with nothing being baked into the topology, like muscles and such. It's all driven by a muscle sim underneath. This is how films normally do it for hero characters. - Characters without simulation. Here you'd bake forms like muscles into the topology. A lot of the time, you'd blend the two a bit, where the torso is usually pretty evenly gridded for sim, while the head is more bespoke topology for wrinkles. For games, it's mostly the same, just more low res :) It's also not subdivided, so you can often do more cheats there. Hope this helps! /Henning
I really like decimating environment things because of the look it gets, very suitable for dirt and rocks. I've been making things straight in maya by using texture deformer with procedural noise/fractals on a high poly object, the reduce by at least 75-80% to make sure it's mostly triangles. I might try it for status, see if it looks chiseled. Strongly dislike having to rig or UV zremeshed things tho, but they're not bad when still sculpting and you're past the dynamesh.
Hey! This is a really interesting way of doing it. We did something similar in a tutorial years back, where you use a texture deformer to get some cool results, and then you triangulate it to death afterwards. Thanks for sharing, Kristine! :) /H
A bit late to this party but one person I met, in this case a miniature figurine sculptor, uses zRemesh then uses tools like Blender with the ReTopoFlow Add-On to retopologize it manually and pose it to where he can make multiple variants then can go in and add some more advanced details to each variant.
For me personally: I prefer decimated meshes for non-organic for gaming only. And ONLY if it's at the final stage after being uv mapped and textured. I'll decimate them when I'm ready to import them into Unity for gaming. If I want it for renders inside Cycles or Arnold, I'll keep the proper quad topo instead. Light seems to bounce off of it better inside these render engines. As for organics, I don't understand why anybody would decimate a face, LOL. Now to watch the video and see if you two enlighten me on anything to the point I'll edit my post :) Edit: K, I was thinking NGons, but I learned tris were bad, so I stay away from them at all costs, lol. Unless they're hidden somewhere away from cameras. Also, I'm not sure what the standard is in production for gaming, but from what I've been doing the past 6 weeks for assets I've been sticking with good topology until my final result where I'll decimate it in Zbrush and copy the UV map over to it, save it as FBX and import it into Unity where I'll make materials for the texture maps, add the materials to the object and compile the result.
This works pretty well for certain cases for sure, but for the most hero characters we'll need to retopo it manually for ultimate control. Like we discussed in the video though, sometimes different pieces are done with manual and some for auto, like if a character has ton of rocks on his back. The main body would be done by hand, while the rocks could be auto. /H
Always marvel at the way DynaMesh decides to use triangles, but love it anyway. Sculptris Pro really makes my skin crawl for the same reason. Both amazing tools. Been modeling in 3D for over 20 years and re-topology just wasn't a thing that anyone thought of then. You had to build it right from the beginning and yes, n-gons were deadly and triangles were useful tools, but never the goal. Needless to say, Zbrush continually blows me away with tools and workflows that don't even seem possible. Makes me feel really old.
Hi, beginner here. Could someone explain what they mean around 4:30 about ZRemeshing --> UV --> Decimation? Wouldn't the decimation break the UV? Maybe I'm missing the point of decimating then...
I guess I misunderstood something about about retopo. Does a dense sculpted mesh get used in render after it’s been retopologized (ie parented to the rigged retopo mesh), or is sculpting only a concept/composition tool?
hey i have a problem...i am making a full character and i want to animate that character. i used zbrush to sculpt the character and now i am making the low poly using the topogun retopology.. so the issue is do i need to do the back side of the clothes tooo??????? if i only did the front side of clothes or any other accessories will it work when i animate it ????????? please help it is hard to do retopology of hidden side of small parts like gloves and boots????
Ok - so (trying to get my head wrapped around this) - do you bake the detail on top of the good topology then (Middle)? Because I can't understand going through extensive detail sculpting and then lose it with low poly afterwards. Clearly there's something I'm missing. And if you do bake the detail like a normal map, how does one do that. [I use Blender.]
Probably a dumb question, but, what is zremeshing (just got zbrush and don't know much about it yet) and decimating? Is having good topology a benefit for portfolio pieces?
Zremesher is a feature found in ZBrush which does the topology in one click. Decimating is drastically reducing the polycount while maintaining the details, by triangulating the mesh. You cant use this for deformation. Good topology is definitely beneficial to have in a portfolio piece :)
Thank you for faving it. :) Though I realize it’s a bit of a mouthful. Just shorten it to “EV”, or why not call it “eevee”, as per the new Blender renderer, or the Pokemon character. :)
Awesome video as usual, theres not ONE bad video from you guys. As for topo, I would ask, what are you looking at? topology of clothing? human body? a gun? is it for game? movie? Its all so weird because i use triangles allll the time when retopoing clothing ( to get those nice wrinkles) but only in certain areas- never on top of joints. ofc all quads for human bodies ( maybe a one single triangle somewhere) you just need to know how the mesh deforms, if it does at all.
How do you deal with stars and triangles causing slight distortion on the surface of the model in sub d? I don't really see any issues with that on the head in the middle where you have stars, but whenever I try and use them on a non-flat surface, the end result is not perfectly smooth.
The best way I've found is to increase the polycount - if it's really low poly and you subD it, you often get artefacts, like you're saying. You can also try using Open SubDivs instead of the regular Catmul Clark algorithm, which might also reduce it. One of the ways is also a bit hacky, but once you have several layers of texture on top, you often cant tell. Hope this helps! /H
"Good topology is dead..." LOL, good luck with telling this to a VFX company hiring TDs, generalists, modelers and riggers. :-D When it comes to deformations and animation in general, good topology is a must, there is nothing else to say here. If you are working as a 3D artist for, say, book covers, illustrations and stills in general, then you might just not bother and scrap the good topology, for it's really not a requirement and you can easily get away with auto-retopo tools. Though if you are working for the automotive industry you will need good topology every now and again, when it comes to UVs and a CAD geometry is not going to cut the purpose (e.g. soft parts like seats, or complex ones like the steering wheel...). So there you have it, you will still be required to have the know how, no matter what. Personally I keep sticking to it because is always a good process that keeps you trained and up to date with the latest technologies and, last but not least, it makes you a much better artist to hiring companies eyes. I am talking by experience of course, not only expressing my opinion here.
What about ZRemeshing with ZRemesher guide brush, i think it can make almost a perfect retopo, then you can correct the topology manually if you let it lowpoly
As a character modeler and rigger, I would say zremesher is far from enough for facial expression. body is probabaly fine. and doing topology is a relaxing process in my book. a day with happy music or backgroup flipnormals stuff lol.
Great discussion. For one thing, retopology follows certain rules and isn't really that creative, it seems possible to be done well by machines. If machine learning gets involved in retopo, auto retopo might actually become good. By then maybe we'll be able to auto retopo the model and only do some minor fixing afterwards, saving a lot of time. The same goes for rigging. Hope there're people working on it and this would come ture soon.
Agreed, I think retopo will be mostly automated in the future once we get tools powered by machine learning. Until then, definitely learn to do proper topo :) /Henning
Can someone please tell me what zeromesh is? I think that's what they're saying. They use the term when refering to the head on the right. I've found nothing on google or chatgpt so i guess theyr'e saying somethign else? edit: nvm i finally found it - I still dont know what they're saying but i think they're talking about: ZRemesher is an automated retopology tool in ZBrush that is used to generate new topology for 3D models. And the link to Danny mac's video is private so we cant see it.
The only time i would use 5 sided faces or more than five sided faces would be on perfectly flat hard surface models or in hardsurfaces areas. As long as there isnt a tiliable crazy texture that needs support from topo
We did one in Maya on doing a head, which might be interesting: ruclips.net/video/9N4rG5qHWgk/видео.html We probably won't do one on ZBrush though, as you never get amazing topology from Zremesher. Maybe we do a tutorial showing the strengths and weaknesses of Zremesher, though!
Cool, I could dig that. Apart from Zremesher's limited sufficient cases where it can be useful, do you tend to believe Zbrush as a whole is subpar/useless for retopo? As in, just ship it over to Maya or start in Maya even to handle all the topology? I have 0 experience with Maya. Are their retopo tools easier or more effective than the zsphere adaptive skin method?
That'd actually be amazing. Beginner question - if I'm doing retopo in ZBrush for enviro assets, would ZRemesher be okay? If so, what would be a general workflow for this? (i.e. do I decimate first then ZRemesh or vice versa?) Thanks for all your content!
it depends on what you want but you can decimate and if the object is not gonna deform or anything you can leave it like that.... you can duplicate your hi poly zremesh it and then get uvs from uv master and then decimate it and then proyect the mesh to you original to keep the form as close as the original (zremesher tends to no care that much about your original form), but i normally dont like the uvs from zrbush..
Would be cool if you'd defined 'Zeevermush' / 'Zebrushmesh' / 'Zevermesh' in the beginning.. Sounds like some Zbrush feature.. Not everyone knows all these things. Edit: Sewermush
Can you guys tell me how I can identify good topology? For me, the one on the right is more subdivided, and that's all I can see. How can I see the difference? I tried retopo my model manually and compare to the remeshed result. I cannot tell which one is better, and I can UV unwrap both of them quite easily. I am confused.
You can totally use ZRemesher AND have good topology. The video shows ZRemesher without any change to the default behavior. You can literally paint your edgeflows before ZRemeshing, and then do a ZRemesh. Works beautifully.
Uh, games, well, my work is pretty low poly so I have to do my topo :P I can use all the triangles I want, I only keep it low because having loops is much easier while working. I will manually add triangles to quads that are folding annoying, happens lots near the brow and mouth, I don't trust my exporter to do it right. Skinny triangles still shade bad even in a static model. I think it took me a year or so to get over the ALL QUADS mindset from school. Triangles are so good. Might be interesting to hear from game artists working with engines that subdivide tho. My advice is keep a few around, it's rarely worth your time chasing down a stubborn triangle.
While it definitely depends, I could tell before you guys even asked the question which model was which with the heads. You could tell mostly by how nice the eyes ended up looking. I think most people can tell at a glance, really.
Since it's mildly related, and it was brought up in the video. As someone who is just starting to see positive returns on his art skills for the first time, can explain to me the cons of using Auto UVing tools in a pipeline that uses Substance for texturing?
Good topology will never die. Truly if anyone wants to know how topology works, create something, texture it, rig it and animate it. Trust me if you don't know good topology then when your are rigging or animating, the reasons will show. By texturing it, the mistakes will be shown more. So for characters, remember your loops and masks (eye mask, ear loop etc) and you will be fine. For environments remember leaves bend, trees sway and rocks crack so just because its not a character doesnt mean to throw topology out the window. That being said. There are parts of even characters that dont need good topology. The better the topology the better the deformation so if it doesnt need to deform that much then the topology matches. For example a snails shell wont deform so technically you can get away with bad topology Personally my methodology makes it so retopoing is easier. People like to start with sculpting, I start with modeling. I build a well made character then head to Zbrush or mudbox or whatever you use for the fine details. Depending on what I do in Zbrush, my retopo time is cut significantly. Plus if I make a destructible mesh for a video game then I would like to control the shape, size and direction the shards go. That is 100% topology driven. Retopoing also makes it so I can plan out how I am going to UV. My cuts are a lot cleaner. I plan out how to rig, I can test the deformation. I plan on how the animation might go. All this will help not only you but if you are part of a team then you will help the rigger and animators too. Outside of things deforming, as long as the silhouette looks like whatever you are making, the topology doesnt matter that much. Bad topology will look ugly, UV ugly, show your lack of skill and lack of care you have for modeling but in the end as long as it works as you intend then nothing else matters. There truly are no rules because you can get to the end hundreds of different ways but there is a difference between have bad topology because you decided that it would cut down on time and still look good vs bad topology because of your lack of knowledge as to why regardless of the end result. Create things the long way and understand why things are the way they are. Then find the shortcuts that we all use.
The idea behind this video is really good and though I like the knowledge mentioned, it's a bit disappointing that with all the descriptions of how important topology is for certain situations, there are no visual examples here of the results achieved with one vs the other when deforming. Everything mentioned is true but the ideal would be to see how a ZRemesher object deforms poorly vs a proper manual retopology, since that is really the time when you can tell the difference the most - maybe it could be shown in a future video? :)
Hi! Thanks a lot for your constructive feedback - it's very helpful. You're spot on here: Showing specific examples regarding when and why you should and shouldn't use ngons, zremeshing etc, would've been a really good idea. Keep in mind though, that this would've taken a lot longer to produce in terms of prep time for the video, and as we aren't making money off our YT channel (tiny bit in ads), we can't spend too much time prepping per video. That said, it's definitely something we'll consider for the future, where we'll try to include more visual examples. Thanks for your comment! :) /Henning
A prominent ZBrush tutor and tech expert said he thinks Manual Topo will soon be fully and accurately automated. Do you guys know of something on the horizon? Also, is ZBrush or 3DCoat better at autoretopo? Thanks!
I also think that will happen in the next 10 years, but right now with the current tools, it can't be replaced. In games and VFX (and more cases, for sure), we simply need perfect control of our topology. The tools we have now aren't smart (they look at the curvature and such of the model), so they can't create topology based on specific use cases - yet. The moment we have some kind of intelligent machine learning tool however, this changes. In the big studios, manual retopo has been moving away gradually though, but that's due to it being outsourced, instead of being automated. I haven't used 3D coat a lot, so honestly I can't give you a good answer here. Regardless, none of them will be miles better than the other as they mostly work based on the same principles. I hope this helps. /Henning
Hmm. It's really about the word : It depends... But most of us are still not in production or not working in the industry. This is where the confusion comes from. Of course if you oversee mechanism then you can decide
The thing with good topology is that you have to ask yourself: which will eat up the least time: making topology by hand or dealing with the problems that could arise from automatic retopology?
That's my position on it too. If you need proper topology, doing it by hand properly is pretty much always more time efficient compared to Zremeshing + fixing it. /H
![Seungmin "그렇게, 천천히, 우리(As we are)" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/kAzmhLHePqU/mqdefault.jpg)
When in doubt ask your local rigger. He will tell you what he needs. If you ever give zremeshed models to your rigger he will appear in your bedroom at 2 AM wailing "Why, oh tell me WHY??"
For sure! The rigger will be under your bed if you give him/her Zremeshed topo.
xDDDDDD
As a rigger, I can attest to this!
@Yu Tubaru I love too if I don't have short time to finish it otherwise just pain in the ass.
You can't call us Riggers, you can say Rigga
Good topology is so important for animation.
it really is. For fully animated characters, you definitely need proper topo.
If you were animating, the heads on the left and right would be no-go areas. Triangles and ngons deform the worst, and too many polygons will make weight painting harder, and make your scene run slower. Middle one gets stars
No Shit
@@SugarCrazedInsomniac the problem with tris and ngons isn't the performance, its how they deform when verts are moved
Hard surface modeling too. If your topology bad it can give bad shade...
So, here's my 2 cents: I really wish companies would invest in AI trained retopologizer, as what is seemingly happening with rendering RnD today - AI trained cloud renderings or GPU-based denoisers to name a couple of examples. No offence to someone who likes it, but it's just a mundane thing that eats too much time as of now.
Not really. You want the control of manual topology, as it allows you where to emphasize or remove detail.
I think tools like this will come in the next 10 years, for sure. Currently most of the retopo in the big studios are done overseas in India and China, as you're right - it is a mundane thing to do and it eats a lot of time (from a studio scheduling point of view).
We already have tools like Wrap3 which makes the job a lot easier when doing humans.
/H
Have you seen this Siggraph vid ruclips.net/video/H8K5CyQB_kc/видео.html. Smart Retopology tools might look something like this, essentially automated retopo tools with a certain level of human input. This gives you the best of both worlds. Plus if you're feeling lazy and the asset doesn't require too much care you can just let the automated tool do the job.
that is insane. That is what Zremesher's descendant should be
I love you guys. Every time I look at your content I learn something, 90% of the time it's just something you say offhand. This time I learned about wrap3. Very cool
Good topology is underrated! I feel like the most underrated aspect being how useful good topology is for sculpting. I sculpt mostly for illustration and retopo manually every time because it's so useful to have those low subdivision levels flowing how you need them to.
good topology is the heart and soul of polygon modeling, which sculpting is 100 percent part of imo :) Totally agree about sculpting on good topology, everything JUST WORKS BETTER.
silly billy
Artificialhype you do realise I'm not opposed to this video?
Artificialhype dude, I did watch the whole thing and the general view is the same as mine. I am in agreement with most of the video. You seem to think I made a video to criticize this one, I quite clearly said I "respectfully disagree" with some points. "Respectfully" because I respect the point of view given and we're essentially on the same side but have a different opinion on a few nuances, and I don't disrespect theirs. You're the only person arguing *for* bad topology here. And it's clear you're unwilling to look at this from any point of view than your own so this is where I end the discussion and wish you good luck in your topological ventures *tips fedora*
Thanks for your input, Danny! Big fan of your work :)
You're spot on - clean topology can be amazingly useful when doing illustration too, as the lower levels allows you to really get a clean base. I find that when doing stylized work, that it's rather easy to get messy shapes in ZBrush, so a good, clean base can help here.
/Henning
First off, great points to bring to light again in an era where technology has improved exponentially.
Of course everyone is entitled to their opinion...from their experiences or what works efficiently for them. If there are "issues" with their retopologized sculpt, what may be a "difficult" fix for some might be an "easy" fix for others depending on their knowledge and needs. I've been in the game industry for over 17 years so I come from the low poly (box modeling first) era where "manual topology" was all you could do. But as with most artists, I have evolved my 3D character workflow. I create models for both games, illustration and sculpts for print nowadays.
Before I jump ahead, I think it is very important to have an understanding of manually retopology of a hi-resolution model, and at times, even certain props, vehicles or environments. It obviously gives you more control to have efficient polygon count, deformation and a better flowing model overall...especially in game engines. But on other hand, I believe that Zbrush gives the artist a freedom...a freedom that they didn't have with other 3D modeling packages; to quickly prototype/conceptualize characters without having to worry initially about deformation, polycount, etc. Zbrush art history has shown that there can be great character sculpts even from pros who haven't manually retopologized (or maybe haven't gotten to it yet).
It seems the bigger argument here is when to do manual retopology. Some prefer it earlier but I have preferred to do it much later towards the end of the pipeline. I give my arguments below.
Auto-retopology was never meant to replace manual retopology. It was meant to get the artist some "almost instant" results while you are sculpting to better flowing topology than it was in its prior state.
It would take you somewhere between 30 minutes to hours manually retopologizing a character (depending how complex it is).
I found that using tricks in auto-retopologizy will get you nice deformations so can to continue sculpt without losing time or worrying about sever topology issues.
Sure it's not perfect but it's very quick and workable enough where its not difficult to manipulate if you run into "topology" quirks.
That's where auto-retopology tricks comes into play to help speed your workflow along.
I think doing manually retopology too early can be counter productive at times because, as we know with clients, there will be many changes to a character...sometimes even drastic where you would have to do manual retopology again. So now you're spending another extra 30 minutes or hours manually retopologizing again. So there are several cases where manual topology treatment is best served until close to the end. Time could be spent better elsewhere down the pipeline. Topology is very important but the client is not going to scrutinize incorrect topology on a character concept or translation when you present it. All that matters in this stage with Zbrush is to quickly get the translation to the 3d model; the likeness or appeal of the character, color, flow, pose, silhouette,...maybe even textures, lighting, etc. As @FlippedNormals showed, you can barely see a difference and manual topology is not as vital when it doesn't matter to the client....or even to artist at this point. Perfect topology is not necessary in this part of preproduction but you still have to have better working topology than you had. That's where knowing how to get nice flowing topology quickly with zremesher will save you a bunch of time.
We know that the results of auto-retopology are still unpredictable even with the current technology but it is a godsend when you know how to use it quickly for your advantage when you need it most. I think the stage of manual retopology for games, films, illustrators and digital sculptors comes down to their preference. I think it comes down to this. If the artists feel they have better control doing manual retopology earlier, then that's the workflow that works best for them. If they feel that manual retopology too early slows them down and like the quicker workflow of tricks using auto-retopology then that perfectly fine too.
Now not retopologizing at all? Not my cup of tea but good luck.
Also, another reason I think it is important to learn proper manual retopology and not just rely on zremesher (or other auto retopology methods). Learning manual retopology (like subdivision/box modeling) will give you a foundation of how the polygons interconnect for proper edge flow and deformation which is important for games and film. Sure it's nice to use the quick tools that will "get you by" but I think it's good as a 3D artist to have a foundation and better understanding of topology. This is not a great example but it would be like a character animator or character illustrator that hasn't learned the foundations/principles of animation or drawing. An understanding of proper manual retopology can keep polygon count efficient too since you have more control.
Also, in regards to characters that require low topology, there are certain "configurations" (poles, ngons, closed edgeloop, linked edge loop, etc) that not only look better when viewed in game engines ( or even animated film).
I remember the joys of turning edges in low polygon modeling to get a smoother looking topology even for static objects and characters.
This applies to non-gaming and non-film digital sculpting too but not as strict as game engines. True it may be less of a worry to manual retopologize, but you can get some workable topology in Zbrush with some auto-retopology trickery.
But like I said, although I opt for manual retopology in the early phase of production..... time constraints usually limit how much time I want to spend on one phase of the pipeline.
In my case, I save the manual retopology for the end. We know the product or character will go through several iterations and even total redesigns that manually topology would be better saved until the very end when it is approved.
It doesn't make any sense spending time manually retopologizing twice when you don't have time to spare. That time could be better spent in posing, texturing, pbr creation, rendering, etc. Nowadays I mainly use the mask/polygroups zremesher edge loop creation trick (I first saw it from a Steve James tutorial). It's not perfect but it's a time saver that I have adopted into my workflow when I don't have time to spare.
Another side point is that I think posing the character as fast as you can in Zbrush is so essential because you can see some possible issues in your model you might not have seen in the T-Pose. Manually retopologizing several layers of meshs (apparel, hair, props, etc.) can get daunting and very time consuming with something that needs a quick turn around to the client. That's where knowledge of the tricks and tools in zbrush can be a time/life saver like t-pose master, polygroups to guide zremesher, and even reproject to dynamesh at times. And even if you end up having a noticeably bad topology in one area but zremesher doesn't help, then just retopologize that area.
I personally wouldn't have any difficulty or loss of much time tweaking portions that didn't deform properly if I didn't do manual retopology at first. ( I'm mainly referring mainly to pre-production models where posing your model quickly for approval is necessary for some clients.) The minor adjustments that I had to tweak with the sculpted pose or modified character was light years faster than the time taken to manually retopologize the whole or even part of a character...not to mention the apparel. If the client requested a major change, redesign or other extreme modification, it would have have to retopologized again anyways, so time spent manual retopologizing would've been wasted. If done properly with zremesher, no client is going to be concerned about imperfect topology and edge loops. So like I said, I'm a big supporter of manual topology I feel it's one of the last phases of the character production pipeline. Thanks for listening to my rant. LOL
Hey! Thank you so much for your very interesting and enlightening comment - it's much appreciated. It's fantastic to hear the opinions of somebody who's been deep in the trenches in the game industry.
My experience in the film industry is very similar to yours, in the way that good topology is really a basis for everything, as it makes life a lot easier. That said, it's not always doable from a time perspective, and then Zremeshing stuff can be a great time saver.
Essentially, use whatever tool gets the job done, without religiously sticking to one technique or tool.
Again, thanks for your elaborate insights! :)
/Henning
Thanks guys. I didn't think I was giving any helpful insight and just rambling away. LOL
This was a great topic you brought up though.
It's getting some constructive and heated reactions. hehe.
I'm sure many more qualified industry pros....unlike myself ( I still have lots to learn) will have much to say on this.
I just found you guys within the past year and your videos and website rock! They are very helpful and insightful as well. Thanks again! :)
Ramble away! You have our permission :D We definitely appreciate other qualified opinions here.
You should have done a small animation of like the mouth moving or blinking eyes to show why you need good topology in like a face.
Good idea. We might do this for the next one.
/H
Please do that!
This right here.
All that talk and they never actually showed an example of it in practice so someone like me who has close to zero idea what they're talking about could actually see it doing its thang.
As a game environment artist, the only time i do manual retopo is if i have very, very few assets to produce. Modern baking and material painting tools are so good it really doesn't matter if your uvs and topo are awful. For many environment assets retopo can take several times longer than the actual asset so its really, really not worth it. I either use custom mesh normals so the mesh holds up on its own without a highpoly or i use automatic retopology. Its a good skill to have but its also a massive production bottleneck for environment art. =/
The topology has a role in making human readable texture maps in regards to island layout and separation, which allows for quick touch-ups in 2D software. So there's that. (Again depending on what workflows are preferred.)
For environment pieces that makes total sense. Good topology is for deformations and shading, but with environment assets that don't deform and whose shading is based on a normal map or custom normals, the best method is whatever is fastest and lowest poly. Agreed 👍
isnt topology completely useless if you're not animating the object?
Its useful, but rarely worth the time. Environment topology tends to be different than character topology, like triangles are totally ok. It saves polycount, makes vertex painting smoother, can remove intersecting geometry which is more expensive to render in games, etc... but none of it makes a big difference and can often take longer than actually making the asset itself.
So would you just use something like decimation master for creating your environment model's topology?
OMG, so cool. Ive waited for such a discussion or video about this exact topic for so long! Thanks again guys. A joy to watch and hear ya. Channel has become really awesome.
Awesome, thanks a lot! :D
In topology edges should follow stretch lines. No more information needed.
Also I love very low poly topo for animation. It's such a nice feeling when in the end your mesh looks almost evactly the same when you apply subdivision as it looked before XD
I'm watching this in the future. From the land of 2020. Thanks, guys!
Good topology is incredibly useful for many aspects of 3d production, but there is always exceptions. Most animators and even some Mograph artists would appreciate good topology when working with models that are rigged or being put through some form of animated motion. Topology defines the model's ability to bend, twist, and sometimes even interact with the surrounding objects in the scene properly. Good topology can also be quite useful when any topological customizations are needed in the workflow. For example, new textures or geometry additions on the fly are much easier to create if the topology is logical and neatly organized on the mesh. Sure we have specialized software now with which retopology or painting without defining UV's is possible, but not everyone has immediate access to these workflows and, in a pinch, it is just so much easier to work with clean geometry and edge flow than messy or unnecissarily dense polygons.
Great video !!!!
I used to spend alot of time building really nice topo, until zremesher came out. For a long time I was of the opinion that topology was dead. That was until I went back and started reworking some of my old models. I found that having nice topology allowed me to stay in lower sub divisions for longer, (especially for areas like eyes/lips or anywhere with alot of folds happening) while still maintaining lots of control on my geometry.
Of course if you are doing hard surface it doesnt matter as much. Now I kind of feel like there are times when zremesher will cut it, but for most of my workflows (character artist) I much prefer to have good topo from the beginning. Plus remeshing can be annoying to get the results you are happy with.
Good video guys! I thought your “center head mesh” had 5-point stars to close to the mouth and then you called in out near the end of the video so I’m glad you did that. I also think your nasal-fold edge is a little high above the nostril, but I know everyone has an opinion on topology, so I make mine based on where I worked and what I see some “top” modelers make. I also know it can change…
I have worked at a company where all the topology had to be quads, and as you said it was because of their UV software. I think it’s good to learn to model in all quads, and then if you’re allowed to use triangles, then you put them in non-deformation areas.
Another thing not mentioned is it good to have stars with more than 5-sided-edges and having a star more than 4 edges on the center-line…?
Again, good video guys and this is a subject I could talk about all day... :-)
Thanks a lot for your input, Roger! I know which company you speak of, and they sure know what they are doing :)
The base mesh we are using here could definitely use with some improvements, like you rightfully point out. It was pretty handy that it had some errors though, as it meant we could use it as an example of how to improve topo.
Good seeing you on our channel, and I'd love your input on our other videos :)
/Henning
lol, who else feels from future. 21:45
It's still an issue. Maybe in 2022
@@StephenHarperRaptagon Judging by the AI boom of 2023, maybe in 2024
Hi there! I am an environment artist and I would love to see a tutorial/lecture about how you guys organize/cleaning up a scene after/during modelling everything. (naming convention, material naming, file naming, etc.)
Thanks a lot!
That's a really good idea! Adding to our list.
Nice discussion, thank guys! I've learned a lot by watching this, topology is really important and automated retopology will never replace manual clean and logically setup topology, I agree. Cheers guys!
Thanks! Im glad you learnt a lot from it.
Voxelized surfaces I find are the best for painting with photo-scanned materials with displacement. You need the geo non pinched edgeloops and even sized faces, otherwise you'll most likely get texture stretching or other artifacts.
have you guys done a video about sculpting for games vs films?
Very interesting and informative. Would doubtlessly love more content of this sort. Thanks, guys!
Cheers Nick!
nice discussion, for a newbie like me into this modeling industry its a really good talk to listen into and know of all this terminologies and such
Awesome, that's really good to hear :)
hello, aspiring animator from the land of 2020. it's still an issue. i just wanted you to know
good to know! xD
And what about the tool Zremesher Guide that we have in Zbrush? Are the results much worse than mapping all of it?
@@Andreopimentel a hobbyist that has been working on 3d models (and having completed like 5 or 7 more models since this comment) i can assure you. remesh seems okay if its a static simple model. "okay" being the operative word. it doesn't seem to be that fantastic. i feel like manually retopologyzing is the way to go almost always. it even opens you to make a subdivision surface modifier and a shrinkwrap (at least in blender)
@@taylorfaith2320 yes, but the zmesher guide tool allows you to select where and how the new topology will be created, have you had any experience with that?
@@Andreopimentel afriad not . keep in mind. Zbrush is still a substantial purchase. and as i said . i'm a hobbyist you might find better answers from another person in these comments.
Good topology is important for a few reasons.
Animation is the most obvious.
But in gamedev, it also acts as an optimization technique. The first is a properly topologized model can use much less geometry while maintaining shape in animation.
The other bit is polygon fans, which is a micro-optimization that allows the GPU to blaze threw geometry without wasting time in swapping states
Isnt poly count not as a big deal anymore though? obviously you cant just have a massively high poly count on any thing,. but within reason? like a full character with 90k tris isn't out of the realm of acceptable for modern engines for PC games. mainly talking bout characters here. in game assets like tables and buildings should be low poly as possible , and modular if possible.
@@temptor7585 isn't it the opposite, you can have as many polys as you want on stationary objects right
FlippedNormals what a channel name!!
thanks for sharing your thoughts and insight on this topic and i completely agree with everything you guys said:). As a fellow modeller (4-6 years so far of learning, and starting to get into more character anatomy/sculpting ) and now some experience in character rigging, i found that lately after I created a face and manually retopolgise it. I tend to start testing blend shapes/skin weights, even though i am not entirely great at animating, to show me if the retoplogising i have done actually works or not.
But this ofc is excessive to do in a production with the amount of time it takes to set up. I just been doing it lately as planning on animating a personal project by myself oneday soon😅
Nice discussion here! If you're gonna animate/deform something, DO MANUAL TOPOLOGY! You want to control where the main loops placed. You want to control where the fold loops are gone be. For all of these and much more, please do manual topology!
You said it bro! :D
Hear Hear! Listen to this man he knows topology!
Indeed! You summed up our video pretty well.
/Henning
yep good vid guys
Dead on... When I am doing animation and characters (which I am often doing) manual topology is KING!! The only time I ever use ZRemesher is for the ears, with carefully placed curves.
Amazing video! I'm a game designer learning blender and was struggling so much with trying to eliminate every tri or pole... this video was like a nice ice tea on a hot day. Thank you SO MUCH!!!
It's not only the rigging. There's one situation where a good topology is imperious, and that's when you want to build the uv map. With bad topology, you will have a hard time to create a proper uv map. But there's a workaround, at least in zbrush- you can remesh, sculpt and get your perfect high res model, and then project it onto the low poly one. All the details will go into the normal map.
I work on games, all the things that you guys say still holds true, except yes we can use triangles anywhere, there is no smooth so no artefact in deformation, and in the engine everything is a triangle anyway. On top of that, because the polycount is limited we sometimes need to define the triangulation of a quad to make sure the silouhette looks smooth.
Cheers!
hi I'm watching from the future, from the land of 2022, and currently watching this while waiting for my quadremesher to finish doing its stuff
Good topology is needed for characters. For hard surface it depends. It depends on if you can optimize the asset by a substantial amount by doing it manually. If you feel like it won't matter, then just Zremesh. Reto is very helpful for texturing things and for controlling texal density.
I use triangles and n-gons a lot. Just subdivide once and in a place where you had a triangle you'll have a tri-edge vert, and where you had a pentagon you'll have a 5 edge vert etc. The main "danger" is concave n-gons. the center of subdivision for those a lot of the time ends up outside of its perimeter and that is the main problem. Quads are just way easier to control and that's why you generally want only quads in your mesh.
The use dictates the mesh via the technical direction, but the crafts person defines the skill everyone needs to work with what they made, and the artist defines the look that needs to stay established. Everyone needs to be a generalist in a way in that you need to be part TD, part crafts person, part artist if you want to work with the orchestra that is animation. Because animation is not one discipline but many under a title. Great discussion but a good follow up referencing this is to cover modern retopo tools, what works, and what doesn't.
Im fed up with these videos about topology when nobody explains how to do good topology. What is the methodology behind it and how can i topologise correctly? And how do you get manual topology to have such perfectly straight loops? If anyone has any resources please help me.
Yes it is. Of course for animation but even when modelling it just makes things easier which means things move faster. As for environment assets, to get those nice smooth highlights run across an object nice topology is a must.
Im learning so much, thank nyou for this channel dudes
Good typology is necessary for two reasons first for your uv texture maps etc as well as animation but it all depends on what the project is really but it's best to make it a habit to have good topology flow on whatever it is...simple as that...
Yes totally agree that good topology will depend of the type of project and what will the model will be used on, but i have an interesting question for you guys, and if you could make a part two with this idea that would be awesome. The question is about topology styles, like you mention video game topology is different obviously than animation, and i see a lot of people with different styles of retopology, what are your thoughts on that topic????
Awesome videos guys!! big fan of yours!!!
Great question! So there are two main things to tackle here:
- Topology which will be simulated. This topology is very even and gridded, with nothing being baked into the topology, like muscles and such. It's all driven by a muscle sim underneath. This is how films normally do it for hero characters.
- Characters without simulation. Here you'd bake forms like muscles into the topology.
A lot of the time, you'd blend the two a bit, where the torso is usually pretty evenly gridded for sim, while the head is more bespoke topology for wrinkles.
For games, it's mostly the same, just more low res :) It's also not subdivided, so you can often do more cheats there.
Hope this helps!
/Henning
Love you guys man!
I was listening to this while retop on Maya. Very well explained BTW
I really like decimating environment things because of the look it gets, very suitable for dirt and rocks. I've been making things straight in maya by using texture deformer with procedural noise/fractals on a high poly object, the reduce by at least 75-80% to make sure it's mostly triangles. I might try it for status, see if it looks chiseled. Strongly dislike having to rig or UV zremeshed things tho, but they're not bad when still sculpting and you're past the dynamesh.
Hey! This is a really interesting way of doing it. We did something similar in a tutorial years back, where you use a texture deformer to get some cool results, and then you triangulate it to death afterwards. Thanks for sharing, Kristine! :)
/H
A bit late to this party but one person I met, in this case a miniature figurine sculptor, uses zRemesh then uses tools like Blender with the ReTopoFlow Add-On to retopologize it manually and pose it to where he can make multiple variants then can go in and add some more advanced details to each variant.
Zremeshing and auto retopo is actually a great use for figurines! There you dont have to deform your model a lot, so it's fantastic for it.
/H
For me personally: I prefer decimated meshes for non-organic for gaming only. And ONLY if it's at the final stage after being uv mapped and textured. I'll decimate them when I'm ready to import them into Unity for gaming.
If I want it for renders inside Cycles or Arnold, I'll keep the proper quad topo instead. Light seems to bounce off of it better inside these render engines.
As for organics, I don't understand why anybody would decimate a face, LOL. Now to watch the video and see if you two enlighten me on anything to the point I'll edit my post :)
Edit: K, I was thinking NGons, but I learned tris were bad, so I stay away from them at all costs, lol. Unless they're hidden somewhere away from cameras.
Also, I'm not sure what the standard is in production for gaming, but from what I've been doing the past 6 weeks for assets I've been sticking with good topology until my final result where I'll decimate it in Zbrush and copy the UV map over to it, save it as FBX and import it into Unity where I'll make materials for the texture maps, add the materials to the object and compile the result.
You can combine the Zremesher and manual topology.
I do it and it makes the retopology much easier.
This works pretty well for certain cases for sure, but for the most hero characters we'll need to retopo it manually for ultimate control. Like we discussed in the video though, sometimes different pieces are done with manual and some for auto, like if a character has ton of rocks on his back. The main body would be done by hand, while the rocks could be auto.
/H
7:32 Just add pointiness and AO components to the shader nodes -- cheaper than extra geometry!
Always marvel at the way DynaMesh decides to use triangles, but love it anyway. Sculptris Pro really makes my skin crawl for the same reason. Both amazing tools. Been modeling in 3D for over 20 years and re-topology just wasn't a thing that anyone thought of then. You had to build it right from the beginning and yes, n-gons were deadly and triangles were useful tools, but never the goal. Needless to say, Zbrush continually blows me away with tools and workflows that don't even seem possible. Makes me feel really old.
2020 here and it's still an issue :(
Hi, beginner here.
Could someone explain what they mean around 4:30 about ZRemeshing --> UV --> Decimation?
Wouldn't the decimation break the UV? Maybe I'm missing the point of decimating then...
Hey! You can click a button in the Decimation Master setting to preserve the UVs. It does a pretty good job.
FlippedNormals I'm silly, thanks so much, your content is amazing!
I guess I misunderstood something about about retopo. Does a dense sculpted mesh get used in render after it’s been retopologized (ie parented to the rigged retopo mesh), or is sculpting only a concept/composition tool?
hey i have a problem...i am making a full character and i want to animate that character.
i used zbrush to sculpt the character
and now i am making the low poly using the topogun retopology..
so the issue is do i need to do the back side of the clothes tooo???????
if i only did the front side of clothes or any other accessories will it work when i animate it ?????????
please help it is hard to do retopology of hidden side of small parts like gloves and boots????
Ok - so (trying to get my head wrapped around this) - do you bake the detail on top of the good topology then (Middle)? Because I can't understand going through extensive detail sculpting and then lose it with low poly afterwards. Clearly there's something I'm missing. And if you do bake the detail like a normal map, how does one do that. [I use Blender.]
i didn't catch the ''getting started with sculpting-zbrush for beginners tutorial'' will it be re-uploaded or is it gone?
The video unfortunately screwed up (our bad!), but it will be up in it's full 59 min glory on Thursday at 4pm GMT! We're really excited for it.
For anyone curious, the Zbrush for Beginners video is now live! :)
ruclips.net/video/_yKGfcp2z3k/видео.html
Is there any software today which might help as it is getting close to 2020 21:45
This video was really helpful, thanks!
Thanks a lot, Chris!
Probably a dumb question, but, what is zremeshing (just got zbrush and don't know much about it yet) and decimating? Is having good topology a benefit for portfolio pieces?
Zremesher is a feature found in ZBrush which does the topology in one click.
Decimating is drastically reducing the polycount while maintaining the details, by triangulating the mesh. You cant use this for deformation.
Good topology is definitely beneficial to have in a portfolio piece :)
FlippedNormals Thank you! Is this process and the rest of the modeling process explained in your game character tutorial that is in your shop?
And what about the tool Zremesher Guide that we have in Zbrush? Are the results much worse than mapping all of it?
19:45 “Extraordinary vertex” is the proper mathematical term.
Thank you for faving it. :)
Though I realize it’s a bit of a mouthful. Just shorten it to “EV”, or why not call it “eevee”, as per the new Blender renderer, or the Pokemon character. :)
Is the manual topology correct? I usually do a loop that goes from forehead to chin. You did forehead to ear and chin to ear
Does the zremesher guide lines change anything to do with this video
Awesome video as usual, theres not ONE bad video from you guys. As for topo, I would ask, what are you looking at? topology of clothing? human body? a gun? is it for game? movie? Its all so weird because i use triangles allll the time when retopoing clothing ( to get those nice wrinkles) but only in certain areas- never on top of joints. ofc all quads for human bodies ( maybe a one single triangle somewhere) you just need to know how the mesh deforms, if it does at all.
Is the Wrap 3 a viable substitute for manual retopology?
How do you deal with stars and triangles causing slight distortion on the surface of the model in sub d? I don't really see any issues with that on the head in the middle where you have stars, but whenever I try and use them on a non-flat surface, the end result is not perfectly smooth.
The best way I've found is to increase the polycount - if it's really low poly and you subD it, you often get artefacts, like you're saying. You can also try using Open SubDivs instead of the regular Catmul Clark algorithm, which might also reduce it.
One of the ways is also a bit hacky, but once you have several layers of texture on top, you often cant tell.
Hope this helps!
/H
Sweet, thank you very much for the reply!
"Good topology is dead..."
LOL, good luck with telling this to a VFX company hiring TDs, generalists, modelers and riggers. :-D
When it comes to deformations and animation in general, good topology is a must, there is nothing else to say here.
If you are working as a 3D artist for, say, book covers, illustrations and stills in general, then you might just not bother and scrap the good topology, for it's really not a requirement and you can easily get away with auto-retopo tools.
Though if you are working for the automotive industry you will need good topology every now and again, when it comes to UVs and a CAD geometry is not going to cut the purpose (e.g. soft parts like seats, or complex ones like the steering wheel...).
So there you have it, you will still be required to have the know how, no matter what.
Personally I keep sticking to it because is always a good process that keeps you trained and up to date with the latest technologies and, last but not least, it makes you a much better artist to hiring companies eyes.
I am talking by experience of course, not only expressing my opinion here.
Completely agree! I really appreciate your balanced and insightful reply here. Thanks!
/H
I don't feel like you watched the video. They pretty much agree with you.
Allen Gingrich
I have watched the entire video, like I always do.
Mine was a thought to underline and reinforce what the video ended up with.
What about ZRemeshing with ZRemesher guide brush, i think it can make almost a perfect retopo, then you can correct the topology manually if you let it lowpoly
hey guys what are your thoughts on diamond quads? they are really helpful when it comes to terminating edges.
Here we are in the land of 2020 and still Good Topology MATTERS.
Oh yes it does!
As a character modeler and rigger, I would say zremesher is far from enough for facial expression. body is probabaly fine. and doing topology is a relaxing process in my book. a day with happy music or backgroup flipnormals stuff lol.
Great discussion. For one thing, retopology follows certain rules and isn't really that creative, it seems possible to be done well by machines. If machine learning gets involved in retopo, auto retopo might actually become good. By then maybe we'll be able to auto retopo the model and only do some minor fixing afterwards, saving a lot of time. The same goes for rigging. Hope there're people working on it and this would come ture soon.
Agreed, I think retopo will be mostly automated in the future once we get tools powered by machine learning. Until then, definitely learn to do proper topo :)
/Henning
That's right :) And even then one should still be able to tell good retopo from bad ones.
This was super helpful and funny as well :D Nice work!
Cool, thanks! Also, thanks for tuning into our stream the other day.
We can even use zremeshner guides right for getting those loops.
and what about if we retopologize with proper zr guides and polygroups, could we achieve the game-ready topo this way?
Can someone please tell me what zeromesh is? I think that's what they're saying. They use the term when refering to the head on the right. I've found nothing on google or chatgpt so i guess theyr'e saying somethign else? edit: nvm i finally found it - I still dont know what they're saying but i think they're talking about: ZRemesher is an automated retopology tool in ZBrush that is used to generate new topology for 3D models. And the link to Danny mac's video is private so we cant see it.
Can you zremesh a scanned mesh to better enable uv mapping?
You could :)
The only time i would use 5 sided faces or more than five sided faces would be on perfectly flat hard surface models or in hardsurfaces areas. As long as there isnt a tiliable crazy texture that needs support from topo
Yes, thats when we use it too, and I think that's a perfectly legit way of using Ngons.
/H
Always Edgeflow around Ringmuscles... simple
Any chance you guys might do an actual retopo walkthrough in Zbrush?
We did one in Maya on doing a head, which might be interesting:
ruclips.net/video/9N4rG5qHWgk/видео.html
We probably won't do one on ZBrush though, as you never get amazing topology from Zremesher. Maybe we do a tutorial showing the strengths and weaknesses of Zremesher, though!
Cool, I could dig that.
Apart from Zremesher's limited sufficient cases where it can be useful, do you tend to believe Zbrush as a whole is subpar/useless for retopo? As in, just ship it over to Maya or start in Maya even to handle all the topology?
I have 0 experience with Maya. Are their retopo tools easier or more effective than the zsphere adaptive skin method?
That'd actually be amazing.
Beginner question - if I'm doing retopo in ZBrush for enviro assets, would ZRemesher be okay? If so, what would be a general workflow for this? (i.e. do I decimate first then ZRemesh or vice versa?)
Thanks for all your content!
it depends on what you want but you can decimate and if the object is not gonna deform or anything you can leave it like that.... you can duplicate your hi poly zremesh it and then get uvs from uv master and then decimate it and then proyect the mesh to you original to keep the form as close as the original (zremesher tends to no care that much about your original form), but i normally dont like the uvs from zrbush..
can you give us sum tips on modeling very hard to model smooth object
Thanks for answering my question!
No problem at all! Hope this helps :)
what about zremesh and reproject all detail on some multiple divided zremesh mesh? It keep the detail with the normal map!
Would be cool if you'd defined 'Zeevermush' / 'Zebrushmesh' / 'Zevermesh' in the beginning.. Sounds like some Zbrush feature..
Not everyone knows all these things.
Edit: Sewermush
They at least named it in the video description, "Zremesh".
Can you guys tell me how I can identify good topology? For me, the one on the right is more subdivided, and that's all I can see. How can I see the difference? I tried retopo my model manually and compare to the remeshed result. I cannot tell which one is better, and I can UV unwrap both of them quite easily. I am confused.
I guess you'll see good topology of your mesh once it doesn't look like crap when you're morphing it?
can you make an entire production like animation Pixar level of production just in Zbrush, or do you have to use other 3d programs along side Zbrush?
You can totally use ZRemesher AND have good topology. The video shows ZRemesher without any change to the default behavior. You can literally paint your edgeflows before ZRemeshing, and then do a ZRemesh. Works beautifully.
Hey there - I know I'm a year late and all, but do you happen to have any resources on how to do that?
My character artist used to give me zremeshed models. Drove me nuts. This was in games.
Oh god...
Uh, games, well, my work is pretty low poly so I have to do my topo :P I can use all the triangles I want, I only keep it low because having loops is much easier while working. I will manually add triangles to quads that are folding annoying, happens lots near the brow and mouth, I don't trust my exporter to do it right. Skinny triangles still shade bad even in a static model. I think it took me a year or so to get over the ALL QUADS mindset from school. Triangles are so good. Might be interesting to hear from game artists working with engines that subdivide tho. My advice is keep a few around, it's rarely worth your time chasing down a stubborn triangle.
While it definitely depends, I could tell before you guys even asked the question which model was which with the heads. You could tell mostly by how nice the eyes ended up looking. I think most people can tell at a glance, really.
Since it's mildly related, and it was brought up in the video. As someone who is just starting to see positive returns on his art skills for the first time, can explain to me the cons of using Auto UVing tools in a pipeline that uses Substance for texturing?
when i weightpaint something (in blender) does every verticyneed to be 100% or can it be more?
I'm afraid I dont understand your question - do you mind elaborating?
Good topology will never die. Truly if anyone wants to know how topology works, create something, texture it, rig it and animate it. Trust me if you don't know good topology then when your are rigging or animating, the reasons will show. By texturing it, the mistakes will be shown more. So for characters, remember your loops and masks (eye mask, ear loop etc) and you will be fine. For environments remember leaves bend, trees sway and rocks crack so just because its not a character doesnt mean to throw topology out the window.
That being said. There are parts of even characters that dont need good topology. The better the topology the better the deformation so if it doesnt need to deform that much then the topology matches. For example a snails shell wont deform so technically you can get away with bad topology
Personally my methodology makes it so retopoing is easier. People like to start with sculpting, I start with modeling. I build a well made character then head to Zbrush or mudbox or whatever you use for the fine details. Depending on what I do in Zbrush, my retopo time is cut significantly.
Plus if I make a destructible mesh for a video game then I would like to control the shape, size and direction the shards go. That is 100% topology driven.
Retopoing also makes it so I can plan out how I am going to UV. My cuts are a lot cleaner. I plan out how to rig, I can test the deformation. I plan on how the animation might go. All this will help not only you but if you are part of a team then you will help the rigger and animators too.
Outside of things deforming, as long as the silhouette looks like whatever you are making, the topology doesnt matter that much. Bad topology will look ugly, UV ugly, show your lack of skill and lack of care you have for modeling but in the end as long as it works as you intend then nothing else matters. There truly are no rules because you can get to the end hundreds of different ways but there is a difference between have bad topology because you decided that it would cut down on time and still look good vs bad topology because of your lack of knowledge as to why regardless of the end result.
Create things the long way and understand why things are the way they are. Then find the shortcuts that we all use.
Hello from the land of 2020!
The idea behind this video is really good and though I like the knowledge mentioned, it's a bit disappointing that with all the descriptions of how important topology is for certain situations, there are no visual examples here of the results achieved with one vs the other when deforming. Everything mentioned is true but the ideal would be to see how a ZRemesher object deforms poorly vs a proper manual retopology, since that is really the time when you can tell the difference the most - maybe it could be shown in a future video? :)
Hi! Thanks a lot for your constructive feedback - it's very helpful.
You're spot on here: Showing specific examples regarding when and why you should and shouldn't use ngons, zremeshing etc, would've been a really good idea. Keep in mind though, that this would've taken a lot longer to produce in terms of prep time for the video, and as we aren't making money off our YT channel (tiny bit in ads), we can't spend too much time prepping per video. That said, it's definitely something we'll consider for the future, where we'll try to include more visual examples.
Thanks for your comment! :)
/Henning
great channel and useful tips. Thanks guys keep it going. I will appreciate it for my Gnomon demoreel :p
Awesome, thanks! :D Best of luck on your reel.
I’m an artist who sculpts only for 3D printing, so this really helps!
Why would mesh topology matter for a 3D printed object, which is obviously not animated?
Zoltán Kürti because if I ever did create an animation model, I would know what to do so it could deform properly.
@@vivianphillips768 oh ok. I tought for some reason you need it for 3d printing.
Zoltán Kürti haha no, 3D printing just needs to be a solid model with a hole at the bottom for draining excess material. Good topology not necessary
A prominent ZBrush tutor and tech expert said he thinks Manual Topo will soon be fully and accurately automated. Do you guys know of something on the horizon? Also, is ZBrush or 3DCoat better at autoretopo? Thanks!
I also think that will happen in the next 10 years, but right now with the current tools, it can't be replaced. In games and VFX (and more cases, for sure), we simply need perfect control of our topology. The tools we have now aren't smart (they look at the curvature and such of the model), so they can't create topology based on specific use cases - yet. The moment we have some kind of intelligent machine learning tool however, this changes.
In the big studios, manual retopo has been moving away gradually though, but that's due to it being outsourced, instead of being automated.
I haven't used 3D coat a lot, so honestly I can't give you a good answer here. Regardless, none of them will be miles better than the other as they mostly work based on the same principles.
I hope this helps.
/Henning
Hmm. It's really about the word : It depends... But most of us are still not in production or not working in the industry. This is where the confusion comes from. Of course if you oversee mechanism then you can decide
Good Topology is needed for animation and lighting
4:40 Do you only have a choice of one decimation algorithm?
docs.blender.org/manual/en/dev/modeling/modifiers/generate/decimate.html
Yes - for decimation in ZBrush, there's just one choice.
the rock wont move but you can still brake it which one is better for that
Do you know any good literature about retopology and UV Mapping?
www.hippydrome.com/
This is fantastic!
The thing with good topology is that you have to ask yourself: which will eat up the least time: making topology by hand or dealing with the problems that could arise from automatic retopology?
That's my position on it too. If you need proper topology, doing it by hand properly is pretty much always more time efficient compared to Zremeshing + fixing it.
/H
isn't good enough to use the zremesherguides brush?
Only in certain cases. If youre doing a hero character, you really need to do the topo by hand, as you dont have enough control otherwise.