BRAVO....finally someone is tackling the HARDEST problem that we architects have to deal. THE TERRAIN. Nothing on this earth is orthogonal. Everything has double curvature. There is curvature, torsion, and all those curvatures you could think on a surface. THIS deserves a series of tutorials for Rhino and Grasshopper. The construction of pavers, sidewalks, ramps and rails are our eternal battle.
@@DesignGoBrr i am a sketchup user, i tried to use Rhino in the past but i found it really hard, now i feel like i can use Rhino because you explain really well and you take your time so i can follow the tutorials haha
Yea, automating it would be awesome. Do you know a good approach to Grasshopper reading of text blocks, associating them to the nearest curve (or curves if that elevation has islands) and then interpolating the heights for all contours that do not have a markup? I tried searching for something similar, but the closest0 approach I found was using Computer Vision + Machine Learning combo and it was all written in C++
I have no idea about Machine Learning. Reading text block is hard (impossible?), block management in rhino is bad and grasshopper doesnt support block. But, for the task of making 3D landscape from 2D contour of CAD file, normally you just need a rectangle boundary, in your example you have the good and clean input, so you flip all curves following 1 curve, then take the end points and sort them, then do the series moving by Z. Just saying for the "normal case". In case of bad input file: complex boundary, or the 2D contour missing some lines, the Z distances between lines are.... not constant(!?). so we have to do manually.
@@vu_derArchitekt I think you described something like I was imagining, in a response above ^^. I was thinking semi-manual, to speed work flow but to keep accuracy (Almost sounds like it should be a tool). Can you see a way to manually draw an intersecting line; select contour lines as a list by intersect with this line; present a prompt/take input param to get a seed elevation + increment; move lines in Z?
I’ve found if you rebuild the contour curves with a higher number of points, turn on control points to make them visible and then select the points instead of the curves, the resulting patch will be more accurate
That's such a strange behaviour. Technically the curve input into patch should produce an "infinite" amount of points for the patch to use, so it's so counter-intuitive. Said that - this is Rhino that we're talking about, most of it is counter-intuitive :D
Not sure what "road steeped terrain" is, but if I had to guess - have 2D curves of a road outline, project them onto a landscape surface, use split to split the landscape with the projected curves, take the part that's split off, extrude it down by let's say 10cm, delete top surface of the extrusion.
Yup, remembered that after finishing recording. Also with any kind of "lofting" style tool - discontinuities in curves is a bad thing, so it's much better to have opening corners filleted.
Hey thanks man, really well timed video. Always enjoy your productions. I was wondering if there's an automated way to tackle Z value reconstruction.. height setting.. of 2D contour lines. I'm surprised there isn't - I'm sure you'd know and be using it! I've been working to extract large landscape data from Geotiff DEM files, via QGIS (where I need to adjust, knit, and trim it) - into Rhino + GH for post processing before constructing physical models. I know Blender can be used here too, but it's not my skillset and there's reasons not to go that way. There seems to be a few paths with R+GH including QGIS Plugins for STL export/import; Pixel-to-point methods for point cloud import, and finally contour export for import and reconstruction. But the videos I've watched all show 3D contour lines already in Rhino, nothing on how to get them over.
Thanks! In regards to using an automated way on moving the contours to their respective Z positions - the short answer is - not possible* . The asterisk is there because it really depends on what kind of starting input you have. Almost all site survey data that we use (north Europe) comes in as AutoCAD 2004 format. Every 4th contour is blocked together with text that specifies its altitude. This kind of info is still not enough to automate the positioning as there are too many variables that can happen in between the "marked" contours. Specially islands and non-natural discontinuities. I think in a few years we will have at least a few Machine Learning tools that will help, but as it is now - nope. Said that - IF you have a reliable GIS data source for the site on which you've working - you can just use the Elk plugin for GH to get all necessary info from it. It's already automated for you then.
@@DesignGoBrr thanks a lot for responding. I can see what you're saying, particularly if data available to you is non-negotiable as AC-2004. In the space of this conversation I've actually cracked my workflow, I'll come back tomorrow and (with the help of the online community) get comfortable with contour > srf/mesh + post processing + colouring + export etc. If it interests you at all, I'm working from local government Lidar DEM files in Geotiff format. Which were imported as a set into QGIS, and had to be levelised by conversion to virtual raster. I'm working with clipped regions from here, to produce contours with the (QGIS-native?) contour tool [here's where my mistake was, I failed to generate Elevation values for the geometry!]. Exporting as .dxf with Z. Imported to Rhino beautifully.. and on into GH etc. Is it feasible for you to use DEM data in your work, in place of the 2D contours provided... or is that not appropriate?
Re-reading your response I understand it's probably higher-accuracy, site specific survey data you're working with. Higher definition than available DEM. Could you partly automate it by drawing a bisect line manually. Then using GH, pick up the intersections (ie. develop a deployable workflow for this). Select lines one by one, using the list of intersections (is possible?), then move the lines up to elevation using a seed value, and a series of elevation increments.
@@yachalupson Oh that sounds like a treat to work with. Unfortunately the rollout of "new tech" is being very... slow in our work environment. There are companies who offer DEM data generation, but they don't do geodesic evaluation (drill probing under the site to see the composition of earth layers), only surface mapping, while the guys who do oldschool autocad drawings, also offer drill-probing and do so at a much cheaper price. So even if we used the DEM data people, we'd still need to use the oldscool autocad people for what's going underneath the site. Once the two variants are proposed, guess which one the client chooses every time.
BRAVO....finally someone is tackling the HARDEST problem that we architects have to deal. THE TERRAIN. Nothing on this earth is orthogonal. Everything has double curvature. There is curvature, torsion, and all those curvatures you could think on a surface.
THIS deserves a series of tutorials for Rhino and Grasshopper. The construction of pavers, sidewalks, ramps and rails are our eternal battle.
Hmm sounds like a fun video - will add it to my to do list!
thank you, I almost forgot about 3D landscape. This was a great refresher
You're very welcome!
Thank you! I usually do everything in gh so that video inspired me not to forget about rhino possibilities :)
Oh you should never forget Rhino, it's so nice to just... model :D
@@DesignGoBrr totally agree :D
Thank you so much. This came just in time for me to do a Site Plan for Class :)
Hehe, glad to hear that this one's gonna be useful.
How about BlendSrf for blending the house to the terrain?
Yea, that would probably also work quite well! Good idea.
I am not even a Rhino user but i love your videos, it makes me wanna try Rhino 😁
I'm really trying not to make this into a rhino channel haha. What kind of program do you use?
@@DesignGoBrr i am a sketchup user, i tried to use Rhino in the past but i found it really hard, now i feel like i can use Rhino because you explain really well and you take your time so i can follow the tutorials haha
@@romyratefiarivony2067 aww I feel good hearing that I help people find the correct path out after they steered into the horrible world of Sketchup~
i think you can move the contour lines to their heights by grasshopper, better than manually moving one by one?
Yea, automating it would be awesome. Do you know a good approach to Grasshopper reading of text blocks, associating them to the nearest curve (or curves if that elevation has islands) and then interpolating the heights for all contours that do not have a markup? I tried searching for something similar, but the closest0 approach I found was using Computer Vision + Machine Learning combo and it was all written in C++
I have no idea about Machine Learning. Reading text block is hard (impossible?), block management in rhino is bad and grasshopper doesnt support block.
But, for the task of making 3D landscape from 2D contour of CAD file, normally you just need a rectangle boundary, in your example you have the good and clean input, so you flip all curves following 1 curve, then take the end points and sort them, then do the series moving by Z. Just saying for the "normal case". In case of bad input file: complex boundary, or the 2D contour missing some lines, the Z distances between lines are.... not constant(!?). so we have to do manually.
Yeap, and most drawings that you get from the goverment are dar from being super clean. That's why I go for a very manual approach
@@vu_derArchitekt I think you described something like I was imagining, in a response above ^^. I was thinking semi-manual, to speed work flow but to keep accuracy (Almost sounds like it should be a tool).
Can you see a way to manually draw an intersecting line; select contour lines as a list by intersect with this line; present a prompt/take input param to get a seed elevation + increment; move lines in Z?
I’ve found if you rebuild the contour curves with a higher number of points, turn on control points to make them visible and then select the points instead of the curves, the resulting patch will be more accurate
That's such a strange behaviour. Technically the curve input into patch should produce an "infinite" amount of points for the patch to use, so it's so counter-intuitive. Said that - this is Rhino that we're talking about, most of it is counter-intuitive :D
How to create road steeped terrain in rhino.? Plz reply
Not sure what "road steeped terrain" is, but if I had to guess - have 2D curves of a road outline, project them onto a landscape surface, use split to split the landscape with the projected curves, take the part that's split off, extrude it down by let's say 10cm, delete top surface of the extrusion.
@@DesignGoBrr I tried this method. But the height of the site is too high. How to make a road like a mountain pass?
from where to download rhino furniture, models and people can u please suggest
I get mine from Chaos Cosmos (comes together with vray)
Great video, Actually sweep does a little better job if you place the cross section curves not at the vertex
Yup, remembered that after finishing recording. Also with any kind of "lofting" style tool - discontinuities in curves is a bad thing, so it's much better to have opening corners filleted.
Notification squad
Notification squad is best squad
Hey thanks man, really well timed video. Always enjoy your productions.
I was wondering if there's an automated way to tackle Z value reconstruction.. height setting.. of 2D contour lines. I'm surprised there isn't - I'm sure you'd know and be using it!
I've been working to extract large landscape data from Geotiff DEM files, via QGIS (where I need to adjust, knit, and trim it) - into Rhino + GH for post processing before constructing physical models.
I know Blender can be used here too, but it's not my skillset and there's reasons not to go that way. There seems to be a few paths with R+GH including QGIS Plugins for STL export/import; Pixel-to-point methods for point cloud import, and finally contour export for import and reconstruction. But the videos I've watched all show 3D contour lines already in Rhino, nothing on how to get them over.
Thanks! In regards to using an automated way on moving the contours to their respective Z positions - the short answer is - not possible* . The asterisk is there because it really depends on what kind of starting input you have. Almost all site survey data that we use (north Europe) comes in as AutoCAD 2004 format. Every 4th contour is blocked together with text that specifies its altitude. This kind of info is still not enough to automate the positioning as there are too many variables that can happen in between the "marked" contours. Specially islands and non-natural discontinuities. I think in a few years we will have at least a few Machine Learning tools that will help, but as it is now - nope. Said that - IF you have a reliable GIS data source for the site on which you've working - you can just use the Elk plugin for GH to get all necessary info from it. It's already automated for you then.
@@DesignGoBrr thanks a lot for responding. I can see what you're saying, particularly if data available to you is non-negotiable as AC-2004.
In the space of this conversation I've actually cracked my workflow, I'll come back tomorrow and (with the help of the online community) get comfortable with contour > srf/mesh + post processing + colouring + export etc.
If it interests you at all, I'm working from local government Lidar DEM files in Geotiff format. Which were imported as a set into QGIS, and had to be levelised by conversion to virtual raster. I'm working with clipped regions from here, to produce contours with the (QGIS-native?) contour tool [here's where my mistake was, I failed to generate Elevation values for the geometry!]. Exporting as .dxf with Z. Imported to Rhino beautifully.. and on into GH etc.
Is it feasible for you to use DEM data in your work, in place of the 2D contours provided... or is that not appropriate?
Re-reading your response I understand it's probably higher-accuracy, site specific survey data you're working with. Higher definition than available DEM.
Could you partly automate it by drawing a bisect line manually. Then using GH, pick up the intersections (ie. develop a deployable workflow for this). Select lines one by one, using the list of intersections (is possible?), then move the lines up to elevation using a seed value, and a series of elevation increments.
@@yachalupson Oh that sounds like a treat to work with. Unfortunately the rollout of "new tech" is being very... slow in our work environment. There are companies who offer DEM data generation, but they don't do geodesic evaluation (drill probing under the site to see the composition of earth layers), only surface mapping, while the guys who do oldschool autocad drawings, also offer drill-probing and do so at a much cheaper price. So even if we used the DEM data people, we'd still need to use the oldscool autocad people for what's going underneath the site. Once the two variants are proposed, guess which one the client chooses every time.
EDIT: It is possible; tested, works.