Sign up for my upcoming 3D Modeling course today: www.onmars3d.com/ Patreon: www.patreon.com/OnMars3D for all my project files, 3d models, bonus videos, Discord and more or just to support this channel 🙂 Discord: discord.gg/vybvJnEeya
Blender has the perfect sub div modelling workflow, as you can keep the supporting edge loops live (non destructively) and use the edge weight data to determine how far apart to have them in sections. Not only that, but you can also transfer that stack easily between models and collections. If you are modeling for a game and require more geometry due to its curved nature, you can turn the Bevel modifier off and let the edge weight hold the shape in the subdiv modifier setting. This way you can have your low/mid poly model right out of your high poly beveled model and have that to bake the bevels onto the other.
@@hectorescobar9450 I'll have to go try it out so I have figure out why. I would have thought that you could also add geometry levels for non-sculpted models.
I see this as important viewing for anyone who's starting to get their feet wet in modelling. I hear all the time about "maintaining quads" or some such, and it's great to finally have an explanation.
Lately I have been stopping to check models in Cyberpunk 2077, especially cars, and, well, so many things are baked on the models and sometimes not with great baking, like visible seams on the tires textures and others, but, of course most people won´t stop to do that and everything works from a reasonable distance. Not that I don´t like the work they put on that game modeling but my admiration for their capacity to simplify so much with great visible results.
That's one of my favorite things to do in games, jump into first person or camera mode and pixel peep all the models to appreciate the details (or find the shortcuts they used)
Modo trained my brain to work with the less amount of polygon that looks good when subD, it's super satisfying to see the model bring to life when subdivided. I don't thing I ever have done a model without subD now, but my work is commercial and not video games
You are one of the best teacher on RUclips for hard surface modeling. Do you plan to do tutorials for texturing and rendering on substance/marmoset/real time engine? A good render has also good textures, materials and good lightning and it is pretty hard to know which are the best practices
Hey Mars! Huge respect for your wisdom as always. For a while now you've been my go to number one guy for who I reccomend to beginners that wish to start 3D modelling. The knowledge you give out for free is always massively appreciated, and just wanted to swing by with a thanks :)
Maybe a video to complement this topic, like an update to industry standards, in which you could show real examples of the current state of the industry in terms of poly count as we are all headed in the direction of generally higher poly counts than just 5 years ago.
That's a great point! I've been talking to AAA game industry friends and they tell me weapons for FPS can easily get over 100k tris with all the mods and attachments. Quite a jump in just a console generation.
At Low Poly Tip 3: Tris for Optimization - the other reason you triangulate your model yourself before the bake phase is because on curvature, especially very low poly projects, the triangle cut will have the potential to change the shading of the surface, which will in turn make your baked normal maps look wrong in those areas. You triangulate before bakes because that is then the surface it is mapping normals too, and you can't rely on the engine to triangulate in the exact same way that you would have beforehand - which as stated, can alter the underlying model's normals that the normal map is working off of.
Always thought that if i made any model and it had tris it would instantly become a model with bad topology, thank you for explaining the way to decide if its good having tris or not when creating a model 🤝.
Awesome video! You have condensed most of the key points of 3D modeling into a quick and easy reference! I am teacher myself and I teach both Maya and Blender modeling, but no matter how many good tips out there, the best way to master 3D modeling is to actually model something and anything and persist at it until you master the tools and techniques.
OH MAN thank you so much for this. I teach a 3D Modeling class and I try to impart this information on the students. Loads of them coming from other people's classes into the portfolio class aren't understanding any of this or when each is appropriate. I was about to have to make one of these myself, but I think you've got a lot better examples to use. This is going to be so helpful. Also helps having someone else also saying what I've been saying because they tend to be like "Uh huh. Okay."
This overall changed my perspective on how i use my topology. Cant say im super good at it yet but i was just throwing tris and nGons along parts of my projects to try to make them look smoother but i also never really used sub-d for any of my projects till recently
I "played" with some 3d modeling software 20-25 years ago and too many polygons meant trouble even to do a still image. Hearing him say to don't care to much to density sparked mixed emotions. Like... like when you compare F1 today and F1 at the times of Airton Senna
Hey man! Thank you for that video it really helps! I'm a hard surface modeler and junior environment artist in the animation industry and I'm currently working on a little game environment to test unreal engine 5 and I come to this issue where I know how to model in the sub-D workflow but I was completely lost when it comes to model in low poly! Your video lights few things up thank you! I'm stil wondering how LOD work exactly, and i'm really interested in your topology class! cheer's!
Glad to hear it, hoping to cover LODs in future videos but it comes down to the distance from the camera. The Level of Detail models are loaded in real-time depending on the distance from the camera, so the engine doesn't have to render more tris than it needs to. UE has an auto LOD generator, so you don't have to do it manually.
@@OnMars3D ah yeah I know what's the principle but it's about the application and how to tackle modelling when it comes to this. But I actually didn't knew UE was doing it automatically ahaha thank you!
what I have learned when it comes to sub-d modelling is that you can model your basic shapes in an almost geometric way and then throw a sub-d modifier in order to get the basic shapes you want. you can then sculpt and bake the details you want on top of that
Some things that are lacking so i will add my own input : When to subdivide or add more polys, in my take its very simple, when your angle is getting steeper and you want a smooth surface then add more polygons, but what about the default cube, as the end result, do we need to subdivide it? no, if we want a cube, adding more vertices will change nothing; just slow down performance and subdividing it will make it smoother by rounding it (if using catmull-clark). 5:08 as for the gun example, remember folks, if your mesh is non deforming then it can be split into multiple meshes, that way, you can probably get that gun to 2K-5K vertices instead of 10K, the payment in game engines that don't do any sorting optimization is that you will pay an extra draw call, its up to you to benchmark it and see if its worth it. A word on the fallacy of "Free / low cost" texture baking such as normal maps, displacement maps and bump maps; A lot of people come under the impression that its "Faster"; no folks, there are NO FREE gifts in life; you just moved the bottleneck from point A to B and maybe B can endure it more; If you ask if you should use normal maps on mobile games; in most cases they will tell you to avoid it; because mobile GPUs can't handle it well. As for PCs, its just the hardware is more powerful. Little tibid for blender users : Normal maps on blender are expansive as hell; that is because half of the work in processing normal maps is done on the CPU instead of the GPU for dependency reasons (Does not apply to video games).
At Sub-d Tip 2: one of the solutions I'd have liked to have seen is that instead of single holding edges either side of a hard edge, two holding lines either side of the border will prevent curvature geometry passing past your hard edge. This is because the curvature is locked into those two quad strips, rather than a single quad strip and then bleeding into the faces it's meant to be holding.
Just posted a pinned comment about this: **I have the flow chart free on my Artstation store here: artstn.co/m/dr3YB Patreon only gives me the option to give assets away for free to paid members so I can't make it free to everyone, apologies about the mixup!
"polycount is not a concern for offline rendering" proceeds to create 2gb model of one single untextured brick. jokes aside, this is a very good video. subbed.
1:46 "Be sure to focus on the funda of quads" IDK.. I think it's the other way around, but then a chicken and egg kind of problem. I only do quads. And i'm doing it for years. But recently, this year, i've worked with a Zbrush god who's super comfortable with ngons. I've learned a lot from working with him. And one of those lessons is that- If i could tell my younger self one of the things i have to master early on - It's definitely going to have to be Ngons. Mastering quads would just cascade by itself.
Just graduated from college. Working on a Game Character portfolio. I thought perhaps characters should all be left in quads when retopologizing, but seems tris are good. When I place renders of the character in the portfolio, should I have one render showing just the quads, and another showing tris? Or do you think it would be better to just have a render of tris except in areas of deformation? I guess faces shouldn't have too many tris either, huh?
My goal has always been to make okay models for games that use low poly models. I test my ability in VRChat, as people with a great range of available performance struggle to load certain worlds and the like, which I like to make; Worked on a world for months that I wanted to be stylized, but need to redo all the mesh, which I have been putting off ':) as it has nothing but n-gons, which look quite bad. Further, you need to optimize the textures and atlas, as most cannot run high resolution stuff well, even if its static and a single 8k texture. Its an interesting challenge that keeps bringing me back to blender and 3d Modeling.
This is a very interesting and informative video with mostly accurate definitions and examples. However, I have a different perspective on the term "SubD modeling." I'm preparing a video to explain why it's more effective to use the terms low-res and high-res modeling with subdivision modifiers. Stay tuned! 😉 Cristian Spagnuolo CG & VFX Supervisor | Trainer, Mentor & Content Creator 🎬✨
Hi! I've seen a lot of 3D modelers (the main example that comes to my mind is chamfer zone) that sometimes use Ngons on their low poly model topology because they know that the export/engine they're exporting to will automatically triangulate those. What's your opinion? is it better to fix by yourself the Ngons or let the triangulation of the program do it's thing? Is there any main difference on the polycount? Love your vids and appreciate your hard work!
I'm glad you brought this up! I plan to cover the remesh workflow in an upcoming video, specifically for game ready assets. It's a great workflow as it allows you to model once with creases, then use remesh to create the high poly. More to come!
Fun fact, a you renders triangles better than quads. You should actually shoot for triangles for speed. The reason new games have low frames? One of them is overusing quads.
ALL polygons are rendered as tris in game/real-time engines. Even if you export your model from your DCC as quads, it will import triangulated in the game engine.
@@OnMars3D all that yap just to agree with me. facepalm. extra work that you could do if you weren't lazy, that the engine doesn't have to do in runtime.
What's your thoughts on using subD with hard soft crease edge in maya. I use it a lot to hold my edges and don't have to use so much edge loops to hold edges. Ends up being a low to mid poly subD
Really Great video, just one doubt, I am a student learning 3d Game art and dev, my teacher says, its illegal (prohobited) to have triangles in a model which is going to be used for games. While here 7:47 you say is opposite, what should I follow? Thanks
Hello, I am Brazilian and I'm starting with Blender. I am looking for a lot of information in this area. I have difficulty dubbing the videos into English, but I try my best to follow your content and dub it into my language. Please recommend older content on your channel or even some courses that I can download to understand Low Poly. Where should I place the vertices and edges appropriately? Please help me.
I assume you're talking about the trigger/hammer and barrel. It's based off the AGL Longcolt from the Trigun anime. There are real revolvers that have the hammer hitting on a lower barrel, you can see an example of this on a Rhino Chiappa revolver.
No, the main things is making sure you have quads in areas of deformation, but the model should not be subdivided in engine. You can, however, texture bake that high poly sub-d detail down to the low poly model, as I explained in the video :)
i really need help reagarding about modelling showreel and what should be the best to make for presonal work or professnial work and how much skill required
i have a question: Were i can find a 3D modeling coure than learn to me this type of nozions? ...Or better, what i need to search, On line course or universitary course, and in the case of the second, what kind of study path is it? Thanks for the attention!😊 P.s. to be clear, im sercing an entire path of study on the 3D modellation, that mean software, algorithm, math, topology, shading, code shading, ecc...
I'm releasing a 3D Modeling Topology Course soon, but you can always look at local institutes for learning 3D Art. Just make sure to ask the students about the program. The institute will always talk it up but the students will tell you the truth :)
i often see some free asset on game asset store use subd method for their free game assets. they mean good but, its too highpoly for simple object but its need much work around to reduce it because they love to use bevel.
I make models for games, but I have never made LPs of models because in UE5, I get nanites that auto-adjust everything. i know it have limitations, plz make a video on it,
Do you mean creasing? Mark Sharp will only "harden" the edge while Creasing will hold the edge during subdivision. It just comes down to the workflow needed for the pipeline.
Correct, for offline rendering, that's more scene optimization. You never want to have millions or billions of tris for no reason. Things like instancing, proxies, standins, help with offline rendering scene optimization.
"We can't just Subdivide our Assets into the Millions and drop them into our games!" The Indie Dev with a Credit Card on the Unity Asset Store: "Watch me."
whats the point of subd modeling hardsurface when you can just use a boolean/cad workflow and bring it into zbrush (or blender remesh/smooth) for polishing? its infinitely faster and more accurate since you can have way more control over bevels and stuff. you can also chain this with subd modeling, but you dont need to mess with the topology of booleans and stuff. its more nondestructive (since the zbrush file maintains history and your pre-zbrush file is super editable) and literally better
I explained why Sub-D modeling is needed for many pipelines like VFX and CGI . You cannot use the remsh workflows for those pipelines. You can however use the remsh workflow for game assets, I'll be covering that in a future video!
@@OnMars3D we are a VR studio (Unity primarily) and have played around with nanite for VR and it works better than expected on PC hardware, lumen is a bigger problem. I think if we give nanite and lumen a few updates to mature it will work very well for VR. Also you get automated LOD and batching that is more effective than classic LOD and batching, basicly the entire scene can be rendered in a few draw calls.
2:29 Not exactly true. You can keep the original model and just use Edge Crease to control how smoothed out the edges will get with Subdivision. This saves you the trouble of evenly slicing the whole cube.
I honestly don't understand why we should learn/improve our 3d skills since 80-90% of the modeling jobs are now moving to India for the fraction of the price, you can't even land a job anymore, not freelance job, neither contract jobs, it's really sad.
It's definitely gotten to a point where working this end of production has tanked. Too many applicants where most of them have no clue how bad they are, 'schools' still cranking out wanna-bes, and work going to mediocre artists for a crappy % of what artists should be paid. The audiences and fans don't care. Shareholders are raping us, fan-boys are flocking to garbage, and idiots think you can steal an asset through AI and your days are golden. (machine learning theft, it's not AI, not even close)
I've lost hope in finding a job for this industry during this past month. Unless someone is willing to move to different cities and work to death until the client is satisfied, then it is pretty hard to make a living our of VFX alone. There are other industries that need our 3d so I will only keep this as a free time hobby thing.
i understand this because i come in 3rd world country with studios here outsourcing and its very competitive. I cant imagine how its like for you guys, i mean im sweating balls alr here the only answer to get jobs in the industry either being a specific niche influenced concept artist, (yes i know its getting harder but generally the skill is very useful), vfx, or like very niche stuff like tech artist or like 2DFX. maybe pixel art, because the demand still going on out there, and other than its a 2d work, mostly is pretty similliar to asset modeler's workflow.
"Before jumping into ngons rabbit hole make sure you understand good topology." - this should be mandatory headline of every beginner tutorial, forum or website. The carving in every free asset offered to public as a reference. There are so many people, especially in Blender communities, who can really mess with newbie's understanding of fundamentals. People repeating like parrots that nothing but quad based topology is allowed, no exception, without actually understanding why is quad only topology recommended in the first place. People who are not even able to imagine that quad is in fact two tris but absolutely convinced about their truth. Or even worse, people, while teach techniques for booleans and ngon topology correctly, lie about the topology rules and purposely withhold the whole truth and the theory behind low poly hard surface modeling. Guys aggressively convincing others that only they are the bearers of truth in the sea of lying assholes and dumb idiots just to lure as much subscribers as possible, promote their crappy courses, always be visible and recommended by YT algorithm. Their business plan works great for them but is seriously harmful and possibly dangerous to anyone new and still confused who finds them at the very beginning of their 3d journey.
Definitely agree, learning the fundamentals of modeling and topology is crucial to being a skilled 3D Artist. If you just boolean and ngon model, you're missing out on workflows for many other industries.
For 3D printing, you're just wasting memory for nothing. Every time i open stuff in sketch, i find these super broken down surfaces... where a simple flat plane is enough. And even as assets, you'll get more than enough with just good textures and a good light + shadow system. Tech is missing the point of KISS, which is very apparent when you look at the tons and tons of games where stuff just doesn't work right and GPUs end up puking their guts out. Light, shadow, texture. That's where it's at. Mastering those and knowing how to form shape for them is superior.
Why would anyone leave Ngons floating about like that at 1:36? You still need a solver to solve to tris, you get no benefit what so ever. Convert to tris and stop flexing a useless "skill" ffs.
1:25 YES. YES, finally someone said it! It's OKAY to have tris and n-gons here and there if you know what you're doing and not going overboard. Yes and they can be problematic on curved surfaces, but you can break even that rule if you, again, know what you're doing. Finally someone to challenge the time-wasting purists.
Sign up for my upcoming 3D Modeling course today: www.onmars3d.com/
Patreon: www.patreon.com/OnMars3D for all my project files, 3d models, bonus videos, Discord and more or just to support this channel 🙂
Discord: discord.gg/vybvJnEeya
Blender has the perfect sub div modelling workflow, as you can keep the supporting edge loops live (non destructively) and use the edge weight data to determine how far apart to have them in sections. Not only that, but you can also transfer that stack easily between models and collections. If you are modeling for a game and require more geometry due to its curved nature, you can turn the Bevel modifier off and let the edge weight hold the shape in the subdiv modifier setting. This way you can have your low/mid poly model right out of your high poly beveled model and have that to bake the bevels onto the other.
Great points, I actually plan to cover this workflow in a future video!
@@OnMars3D great stuff in your channel btw man. Definitely one that f the better ones out there
@@hectorescobar9450 why not use the multi- resolution modifier?
@@BGW0 multi resolution is mainly used for sculpting. Not great for hard surface clean models.
@@hectorescobar9450 I'll have to go try it out so I have figure out why. I would have thought that you could also add geometry levels for non-sculpted models.
I see this as important viewing for anyone who's starting to get their feet wet in modelling. I hear all the time about "maintaining quads" or some such, and it's great to finally have an explanation.
Glad you found it helpful!
I am learning blender as hobby and fun, and yet i have learnt more then with some of my jobs! Thanks for the video
Lately I have been stopping to check models in Cyberpunk 2077, especially cars, and, well, so many things are baked on the models and sometimes not with great baking, like visible seams on the tires textures and others, but, of course most people won´t stop to do that and everything works from a reasonable distance. Not that I don´t like the work they put on that game modeling but my admiration for their capacity to simplify so much with great visible results.
That's one of my favorite things to do in games, jump into first person or camera mode and pixel peep all the models to appreciate the details (or find the shortcuts they used)
@@OnMars3D I do this all the time, I have an unhealthy obsession with inspecting other peoples work while I'm gaming with friends and they hate it :)
Modo trained my brain to work with the less amount of polygon that looks good when subD, it's super satisfying to see the model bring to life when subdivided.
I don't thing I ever have done a model without subD now, but my work is commercial and not video games
You gotta try Plasticity my dude
really clear, short and well laid out. thank you
Happy to share!
You are one of the best teacher on RUclips for hard surface modeling. Do you plan to do tutorials for texturing and rendering on substance/marmoset/real time engine? A good render has also good textures, materials and good lightning and it is pretty hard to know which are the best practices
A good model also has those features you mentioned, yes, but it totally depends on what your end-goal is. :)
@@nanshire2278 Portfolio and in engine
Glad to hear it and yes! I plan to make more tutorials on the materials, textures, and rendering in future videos and courses, so stay tuned!
Hey Mars! Huge respect for your wisdom as always. For a while now you've been my go to number one guy for who I reccomend to beginners that wish to start 3D modelling. The knowledge you give out for free is always massively appreciated, and just wanted to swing by with a thanks :)
Much appreciated, I'm all about sharing knowledge!
Maybe a video to complement this topic, like an update to industry standards, in which you could show real examples of the current state of the industry in terms of poly count as we are all headed in the direction of generally higher poly counts than just 5 years ago.
That's a great point! I've been talking to AAA game industry friends and they tell me weapons for FPS can easily get over 100k tris with all the mods and attachments. Quite a jump in just a console generation.
holy shit trigun. I already modelled this gun years ago, its that cool of a design, and the series itself is fantastic
Another great video and great reference for the basics. Saving this one for sure.
At Low Poly Tip 3: Tris for Optimization - the other reason you triangulate your model yourself before the bake phase is because on curvature, especially very low poly projects, the triangle cut will have the potential to change the shading of the surface, which will in turn make your baked normal maps look wrong in those areas. You triangulate before bakes because that is then the surface it is mapping normals too, and you can't rely on the engine to triangulate in the exact same way that you would have beforehand - which as stated, can alter the underlying model's normals that the normal map is working off of.
Yeah, great point, you should always triangulate the model before baking the same way you would import into engine.
Always thought that if i made any model and it had tris it would instantly become a model with bad topology, thank you for explaining the way to decide if its good having tris or not when creating a model 🤝.
Glad it helped your understanding of topology!
great video, covering all the intricate details, and not just surface level stuff🔥🔥🔥
Appreciate all the help you've given me from your mentorship so far 🙏
Glad you joined, you're showing great improvements!
Awesome video! You have condensed most of the key points of 3D modeling into a quick and easy reference! I am teacher myself and I teach both Maya and Blender modeling, but no matter how many good tips out there, the best way to master 3D modeling is to actually model something and anything and persist at it until you master the tools and techniques.
Appreciate it, and I definitely agree, the best way to learn is just to get out there and do it, over and over again :)
OH MAN thank you so much for this. I teach a 3D Modeling class and I try to impart this information on the students. Loads of them coming from other people's classes into the portfolio class aren't understanding any of this or when each is appropriate. I was about to have to make one of these myself, but I think you've got a lot better examples to use. This is going to be so helpful. Also helps having someone else also saying what I've been saying because they tend to be like "Uh huh. Okay."
So glad to hear that! I deal with the same thing, even live in labs I still get the blank head nods XD
The editing is funky af
I love it
the information you give is really informative! this helps a lot of people deciding suitable workflow down the line! thank you so much!
Glad you found it helpful!
I learn a lot every time when I watch your videos, u answer some of the questions that I search a lot to find the answer. amazing
Thank you my friend, glad you found it helpful!
This overall changed my perspective on how i use my topology. Cant say im super good at it yet but i was just throwing tris and nGons along parts of my projects to try to make them look smoother but i also never really used sub-d for any of my projects till recently
I "played" with some 3d modeling software 20-25 years ago and too many polygons meant trouble even to do a still image. Hearing him say to don't care to much to density sparked mixed emotions. Like... like when you compare F1 today and F1 at the times of Airton Senna
Man I can't wait for your topology masterclass course!
This is exactly what i need! Thank you!!
Happy to share!
Hey man!
Thank you for that video it really helps!
I'm a hard surface modeler and junior environment artist in the animation industry and I'm currently working on a little game environment to test unreal engine 5 and I come to this issue where I know how to model in the sub-D workflow but I was completely lost when it comes to model in low poly!
Your video lights few things up thank you!
I'm stil wondering how LOD work exactly, and i'm really interested in your topology class!
cheer's!
Glad to hear it, hoping to cover LODs in future videos but it comes down to the distance from the camera. The Level of Detail models are loaded in real-time depending on the distance from the camera, so the engine doesn't have to render more tris than it needs to. UE has an auto LOD generator, so you don't have to do it manually.
@@OnMars3D ah yeah I know what's the principle but it's about the application and how to tackle modelling when it comes to this. But I actually didn't knew UE was doing it automatically ahaha thank you!
what I have learned when it comes to sub-d modelling is that you can model your basic shapes in an almost geometric way and then throw a sub-d modifier in order to get the basic shapes you want. you can then sculpt and bake the details you want on top of that
ohhhh the 45 long colt. Vash gonna love this
Great video! You explain things well. Thank you.
Some things that are lacking so i will add my own input :
When to subdivide or add more polys, in my take its very simple, when your angle is getting steeper and you want a smooth surface then add more polygons, but what about the default cube, as the end result, do we need to subdivide it? no, if we want a cube, adding more vertices will change nothing; just slow down performance and subdividing it will make it smoother by rounding it (if using catmull-clark).
5:08 as for the gun example, remember folks, if your mesh is non deforming then it can be split into multiple meshes, that way, you can probably get that gun to 2K-5K vertices instead of 10K, the payment in game engines that don't do any sorting optimization is that you will pay an extra draw call, its up to you to benchmark it and see if its worth it.
A word on the fallacy of "Free / low cost" texture baking such as normal maps, displacement maps and bump maps;
A lot of people come under the impression that its "Faster"; no folks, there are NO FREE gifts in life; you just moved the bottleneck from point A to B and maybe B can endure it more; If you ask if you should use normal maps on mobile games; in most cases they will tell you to avoid it; because mobile GPUs can't handle it well.
As for PCs, its just the hardware is more powerful.
Little tibid for blender users : Normal maps on blender are expansive as hell; that is because half of the work in processing normal maps is done on the CPU instead of the GPU for dependency reasons (Does not apply to video games).
you really stepped further with the editing, its so much nicer.
Glad you noticed, working with an Editor now so we're trying to level up the content!
At Sub-d Tip 2: one of the solutions I'd have liked to have seen is that instead of single holding edges either side of a hard edge, two holding lines either side of the border will prevent curvature geometry passing past your hard edge. This is because the curvature is locked into those two quad strips, rather than a single quad strip and then bleeding into the faces it's meant to be holding.
Great point, I like to add extra segments on my bevels for this exact reason
Thank you!! This is so helpful :)
Happy to help!
Brother, thanks, i actually trying to use subd modeling and more knowledge it's always good.
You are most welcome
this video come at the perfect moment in my life
Glad to hear it!
Trigun AGL Arms attracted more than the title itself
YES FINALLY SOMEONE SEES WHAT IM SEEING
One of my favorite animes 🙂
Thanks for the flow chart man, appreciate it
Just posted a pinned comment about this:
**I have the flow chart free on my Artstation store here: artstn.co/m/dr3YB
Patreon only gives me the option to give assets away for free to paid members so I can't make it free to everyone, apologies about the mixup!
Oh okay, that makes more sense. Ill edit the comment
@@Legit_SuperFall all good!
@@OnMars3D hope you didn't take it personally
@@Legit_SuperFall No worries, I try not to take things personal on the internet :)
Really amazing video! Thank you.
Glad you liked it!
"polycount is not a concern for offline rendering"
proceeds to create 2gb model of one single untextured brick.
jokes aside, this is a very good video. subbed.
Ha, thank you enojyed it!
Thank you!
You bet!
When's the course going to release? Is it applicable for Blender?
Amasing video! Thx a lot!)
Glad you liked it!
Can't wait for this Masterclass! How close it is to us? 1 - 2 weeks?
Not quite that quick, but my plan is to have it ready by the fall!
Awesome, thank you. Can't wait to see what we gonna model :) I really love your approach to the modeling and you explain the principles@@OnMars3D
@@meLaNiXonS No problem, appreciate the kind words!
1:46 "Be sure to focus on the funda of quads"
IDK.. I think it's the other way around, but then a chicken and egg kind of problem.
I only do quads. And i'm doing it for years.
But recently, this year, i've worked with a Zbrush god who's super comfortable with ngons. I've learned a lot from working with him. And one of those lessons is that-
If i could tell my younger self one of the things i have to master early on - It's definitely going to have to be Ngons.
Mastering quads would just cascade by itself.
Just graduated from college. Working on a Game Character portfolio. I thought perhaps characters should all be left in quads when retopologizing, but seems tris are good. When I place renders of the character in the portfolio, should I have one render showing just the quads, and another showing tris? Or do you think it would be better to just have a render of tris except in areas of deformation? I guess faces shouldn't have too many tris either, huh?
If you retopo it, just show that version before it goes into engine (quads and tris). Showing both before and after engine is completely fine.
Amazing video
Appreciate it, took a lot of work!
My goal has always been to make okay models for games that use low poly models. I test my ability in VRChat, as people with a great range of available performance struggle to load certain worlds and the like, which I like to make; Worked on a world for months that I wanted to be stylized, but need to redo all the mesh, which I have been putting off ':) as it has nothing but n-gons, which look quite bad. Further, you need to optimize the textures and atlas, as most cannot run high resolution stuff well, even if its static and a single 8k texture. Its an interesting challenge that keeps bringing me back to blender and 3d Modeling.
great video!
Superb video 👌
Thank you!
This is a very interesting and informative video with mostly accurate definitions and examples. However, I have a different perspective on the term "SubD modeling." I'm preparing a video to explain why it's more effective to use the terms low-res and high-res modeling with subdivision modifiers.
Stay tuned! 😉
Cristian Spagnuolo
CG & VFX Supervisor | Trainer, Mentor & Content Creator 🎬✨
Hi! I've seen a lot of 3D modelers (the main example that comes to my mind is chamfer zone) that sometimes use Ngons on their low poly model topology because they know that the export/engine they're exporting to will automatically triangulate those. What's your opinion? is it better to fix by yourself the Ngons or let the triangulation of the program do it's thing? Is there any main difference on the polycount? Love your vids and appreciate your hard work!
I'm glad you brought this up! I plan to cover the remesh workflow in an upcoming video, specifically for game ready assets. It's a great workflow as it allows you to model once with creases, then use remesh to create the high poly. More to come!
Fun fact, a you renders triangles better than quads. You should actually shoot for triangles for speed. The reason new games have low frames? One of them is overusing quads.
ALL polygons are rendered as tris in game/real-time engines. Even if you export your model from your DCC as quads, it will import triangulated in the game engine.
@@OnMars3D so you admit quads are ruining games. thanks dumbass.
@@OnMars3D all that yap just to agree with me. facepalm. extra work that you could do if you weren't lazy, that the engine doesn't have to do in runtime.
I dont do 3D or anything close to that but that was real interesting lmao great content !!
What's your thoughts on using subD with hard soft crease edge in maya. I use it a lot to hold my edges and don't have to use so much edge loops to hold edges. Ends up being a low to mid poly subD
Where can I find those n-gon wizard post?
Really Great video, just one doubt,
I am a student learning 3d Game art and dev, my teacher says, its illegal (prohobited) to have triangles in a model which is going to be used for games. While here 7:47 you say is opposite, what should I follow?
Thanks
Tris are fine to use, especially for game assets. You just want to keep quads in areas of deformation.
@@OnMars3D Thanks man, That means a lot! 😃
Hello, I am Brazilian and I'm starting with Blender. I am looking for a lot of information in this area. I have difficulty dubbing the videos into English, but I try my best to follow your content and dub it into my language. Please recommend older content on your channel or even some courses that I can download to understand Low Poly. Where should I place the vertices and edges appropriately? Please help me.
I have a lot of older content on my channel, but most of it is Maya. I'll be putting more Blender (and Maya) content out soon!
Where can we find that paper on topology seen @1:38?
nice to see Vash's Gun
Great information and very concise. Though whoever modelled that revolver gun needs to have their eyes checked😃
I assume you're talking about the trigger/hammer and barrel. It's based off the AGL Longcolt from the Trigun anime. There are real revolvers that have the hammer hitting on a lower barrel, you can see an example of this on a Rhino Chiappa revolver.
Is sub-d practical for game models? I've always thought to stay away because of the high resolution.
No, the main things is making sure you have quads in areas of deformation, but the model should not be subdivided in engine. You can, however, texture bake that high poly sub-d detail down to the low poly model, as I explained in the video :)
i really need help reagarding about modelling showreel and what should be the best to make for presonal work or professnial work and how much skill required
Feel free to join my Discord and post there: discord.com/invite/83MMpPrE
Question: when will you release your course? Because i don’t want to buy another course .. thanks
Planning to have it ready by the fall!
you should talk about LODs too tho
i have a question:
Were i can find a 3D modeling coure than learn to me this type of nozions?
...Or better, what i need to search, On line course or universitary course, and in the case of the second, what kind of study path is it?
Thanks for the attention!😊
P.s.
to be clear, im sercing an entire path of study on the 3D modellation, that mean software, algorithm, math, topology, shading, code shading, ecc...
I'm releasing a 3D Modeling Topology Course soon, but you can always look at local institutes for learning 3D Art. Just make sure to ask the students about the program. The institute will always talk it up but the students will tell you the truth :)
8:50 cant find this thing anywhere on patreon..
Unfortunately, I can't put free content on Patreon unless you're a member. However, I uploaded it here: artstn.co/m/dr3YB
I've mastered mid level polys, cause its all my work requries lol.
Yeah, I get that, that's why I showed a higher range for the revolver, around 30k with extra detail that isn't in the much lower version.
i often see some free asset on game asset store use subd method for their free game assets. they mean good but, its too highpoly for simple object but its need much work around to reduce it because they love to use bevel.
I make models for games, but I have never made LPs of models because in UE5, I get nanites that auto-adjust everything. i know it have limitations, plz make a video on it,
What about endgon workflow? or otherwise called as triangulate method.
Bro that spongebob meke me lought so much.
It was one of my favorite memes in the video 😂
Can't one use Marked Sharps to avoid the problem of Sub-D Modeling degrading sharp edges?
Do you mean creasing? Mark Sharp will only "harden" the edge while Creasing will hold the edge during subdivision. It just comes down to the workflow needed for the pipeline.
the Gigachad of blender
😂
optimization matters even if you're rendering with a normal renderer
Correct, for offline rendering, that's more scene optimization. You never want to have millions or billions of tris for no reason. Things like instancing, proxies, standins, help with offline rendering scene optimization.
The gigachad of 3D
😂
"We can't just Subdivide our Assets into the Millions and drop them into our games!"
The Indie Dev with a Credit Card on the Unity Asset Store: "Watch me."
I've been there, some UE and Unity Assets are a mess!
whats the point of subd modeling hardsurface when you can just use a boolean/cad workflow and bring it into zbrush (or blender remesh/smooth) for polishing? its infinitely faster and more accurate since you can have way more control over bevels and stuff. you can also chain this with subd modeling, but you dont need to mess with the topology of booleans and stuff. its more nondestructive (since the zbrush file maintains history and your pre-zbrush file is super editable) and literally better
I explained why Sub-D modeling is needed for many pipelines like VFX and CGI . You cannot use the remsh workflows for those pipelines. You can however use the remsh workflow for game assets, I'll be covering that in a future video!
Quick question support edges or creases?
Creases first, once form is final, then support/holding lines
Make it tutorial game workflow modeling please, (highpoly,low poly, baking)
I hope this doesn't age like fine wine 😭🙏
Which software are you going to use in your course?
With nanite yuu can use the sub d model directly in the engine
For high-end target hardware, sure. But not for more optimized hardware like VR and Mobile.
@@OnMars3D we are a VR studio (Unity primarily) and have played around with nanite for VR and it works better than expected on PC hardware, lumen is a bigger problem. I think if we give nanite and lumen a few updates to mature it will work very well for VR. Also you get automated LOD and batching that is more effective than classic LOD and batching, basicly the entire scene can be rendered in a few draw calls.
@@andersmalmgren6528 That's good to hear. With every UE update, Nanite and Lumen are becoming more optimized.
2:29 Not exactly true. You can keep the original model and just use Edge Crease to control how smoothed out the edges will get with Subdivision. This saves you the trouble of evenly slicing the whole cube.
Dive into an exclusive interview with Binance's CEO for a peek into future developments
"If your goal is to do offline rendering, you dont need to worry about polycount" idk abt that chief
Within reason of course, scene optimization for offline rendering is important too.
vr complicates things even more. it beneits vastly from texture quality and poly count yet due to rendering two unique view points at once.
What are you pc specs?
AMD 7950x, RTX 3090, 64 GB RAM, 4TB SSD NVME
I honestly don't understand why we should learn/improve our 3d skills since 80-90% of the modeling jobs are now moving to India for the fraction of the price, you can't even land a job anymore, not freelance job, neither contract jobs, it's really sad.
It's definitely gotten to a point where working this end of production has tanked. Too many applicants where most of them have no clue how bad they are, 'schools' still cranking out wanna-bes, and work going to mediocre artists for a crappy % of what artists should be paid. The audiences and fans don't care. Shareholders are raping us, fan-boys are flocking to garbage, and idiots think you can steal an asset through AI and your days are golden. (machine learning theft, it's not AI, not even close)
I've lost hope in finding a job for this industry during this past month. Unless someone is willing to move to different cities and work to death until the client is satisfied, then it is pretty hard to make a living our of VFX alone.
There are other industries that need our 3d so I will only keep this as a free time hobby thing.
the tutorial is for indians
@@vfxartdsInDiA bAD aMeRiCa GoOd🤡
i understand this because i come in 3rd world country with studios here outsourcing and its very competitive. I cant imagine how its like for you guys, i mean im sweating balls alr here
the only answer to get jobs in the industry either being a specific niche influenced concept artist, (yes i know its getting harder but generally the skill is very useful), vfx, or like very niche stuff like tech artist or like 2DFX.
maybe pixel art, because the demand still going on out there, and other than its a 2d work, mostly is pretty similliar to asset modeler's workflow.
"Before jumping into ngons rabbit hole make sure you understand good topology." - this should be mandatory headline of every beginner tutorial, forum or website. The carving in every free asset offered to public as a reference.
There are so many people, especially in Blender communities, who can really mess with newbie's understanding of fundamentals. People repeating like parrots that nothing but quad based topology is allowed, no exception, without actually understanding why is quad only topology recommended in the first place. People who are not even able to imagine that quad is in fact two tris but absolutely convinced about their truth.
Or even worse, people, while teach techniques for booleans and ngon topology correctly, lie about the topology rules and purposely withhold the whole truth and the theory behind low poly hard surface modeling. Guys aggressively convincing others that only they are the bearers of truth in the sea of lying assholes and dumb idiots just to lure as much subscribers as possible, promote their crappy courses, always be visible and recommended by YT algorithm. Their business plan works great for them but is seriously harmful and possibly dangerous to anyone new and still confused who finds them at the very beginning of their 3d journey.
Definitely agree, learning the fundamentals of modeling and topology is crucial to being a skilled 3D Artist. If you just boolean and ngon model, you're missing out on workflows for many other industries.
Grade A Meme game
Trying to have some fun with it!
I was JUST making a trigun revolver!
YES
For 3D printing, you're just wasting memory for nothing. Every time i open stuff in sketch, i find these super broken down surfaces... where a simple flat plane is enough. And even as assets, you'll get more than enough with just good textures and a good light + shadow system.
Tech is missing the point of KISS, which is very apparent when you look at the tons and tons of games where stuff just doesn't work right and GPUs end up puking their guts out. Light, shadow, texture. That's where it's at. Mastering those and knowing how to form shape for them is superior.
why does he look like he's a something created in blender
It's all about the lighting 😎
Why would anyone leave Ngons floating about like that at 1:36?
You still need a solver to solve to tris, you get no benefit what so ever.
Convert to tris and stop flexing a useless "skill" ffs.
1:25 YES. YES, finally someone said it! It's OKAY to have tris and n-gons here and there if you know what you're doing and not going overboard. Yes and they can be problematic on curved surfaces, but you can break even that rule if you, again, know what you're doing. Finally someone to challenge the time-wasting purists.
I don’t care