A few notes and questions: 1. How about metallic paint with pearls instead of flakes? I personally prefer flakes with a realistic scale to them over white noise. 2. For flakes in UV tangent space, shouldn't color output be [0, 0, 1] to [ 1, 1, 1]? Normal maps in this space always have blue (almost) maxed at all times. I use full color but in object space, although it will fail too if the effect is too much. 3. For paint, have you tried a layered approach to add depth and color variations? Instead of a single paint and coat, the paint consists of several glossys with different colors, roughnesses, and even normal response. Similar to how you'd do "hazy gloss". A more advanced version would be to use voronoi (color->separate color.hue, maybe to a power to shape the curve) to select the parameters that goes into a single glossy, which now results in a single shader call per pixel sample and thus more efficient. 4. For carbon fiber, I like the single parameter to control the (simple) weave pattern, much better than mine. 5. Not a fan of calculating normal maps when bumps are much more intuitive to control, and are not affected by the tangent space problems with triplanar (sharp like yours, or smooth) mapping. I'm a big user of smooth triplanar mapping for other purposes, but I can't imagine using them on carbon fiber car parts. In manufacturing, it's literally a sheet of carbon fiber weave you press onto the shape, so UVs just make a lot more sense. You also get tangent space UVs for anisotropic shading (tangent node), which is my preferred method. A thing to be aware of is that if you scale the fiber density, you have to scale by 1/scale into bump distance to get the same normals, whereas for normal maps the normals are scale independent. 6. I'm guessing no onion peeling on the coat is a deliberate choice? 7. I would probably still add a little large scale bump on the paint coat to make it less perfect, not sure how perfect fiber resin is. 8. With thick coats there may also be a darkening effect on the substrate; grazing angles being darker than head on angles.
Awesome work! Just a suggestion for your white noise colour generator (18:30) instead of a random cartesian generator, use polar coordinates - i.e. random vector length and random angle (0-360°) - then resolve these back to cartesian coordinates for the next step with a X=vector length*cos(angle) and Y=vector length*sin(angle).
Awesome content. Love that Blender is growing more and more as a tool in the engineering design and visualisation space. Getting an insight into the workflows and thought processes of high-end professionals like this is gold. BCon LA have definitely laid down the gauntlet for quality, for the Blender Conference in Amsterdam to meet in October.
Awesome job, Odi! I would also love to see how you would create a forged carbon fiber shader. Similar to the one found on BMW Vision M Next. Thank you and keep up the good work!
Question: When I apply a shader and then modify its settings like color or scale, it appears to save all these changes to the base material in the library so now every time I want to go back to that base material its now changed. How do I lock the library materials so they wont change?
My mind is blown that you can make make carbon textures without needing to uv unwrap!
Also, this video is incredibly well articulated. Great work
A few notes and questions:
1. How about metallic paint with pearls instead of flakes? I personally prefer flakes with a realistic scale to them over white noise.
2. For flakes in UV tangent space, shouldn't color output be [0, 0, 1] to [ 1, 1, 1]? Normal maps in this space always have blue (almost) maxed at all times. I use full color but in object space, although it will fail too if the effect is too much.
3. For paint, have you tried a layered approach to add depth and color variations? Instead of a single paint and coat, the paint consists of several glossys with different colors, roughnesses, and even normal response. Similar to how you'd do "hazy gloss". A more advanced version would be to use voronoi (color->separate color.hue, maybe to a power to shape the curve) to select the parameters that goes into a single glossy, which now results in a single shader call per pixel sample and thus more efficient.
4. For carbon fiber, I like the single parameter to control the (simple) weave pattern, much better than mine.
5. Not a fan of calculating normal maps when bumps are much more intuitive to control, and are not affected by the tangent space problems with triplanar (sharp like yours, or smooth) mapping. I'm a big user of smooth triplanar mapping for other purposes, but I can't imagine using them on carbon fiber car parts. In manufacturing, it's literally a sheet of carbon fiber weave you press onto the shape, so UVs just make a lot more sense. You also get tangent space UVs for anisotropic shading (tangent node), which is my preferred method. A thing to be aware of is that if you scale the fiber density, you have to scale by 1/scale into bump distance to get the same normals, whereas for normal maps the normals are scale independent.
6. I'm guessing no onion peeling on the coat is a deliberate choice?
7. I would probably still add a little large scale bump on the paint coat to make it less perfect, not sure how perfect fiber resin is.
8. With thick coats there may also be a darkening effect on the substrate; grazing angles being darker than head on angles.
Awesome work! Just a suggestion for your white noise colour generator (18:30) instead of a random cartesian generator, use polar coordinates - i.e. random vector length and random angle (0-360°) - then resolve these back to cartesian coordinates for the next step with a X=vector length*cos(angle) and Y=vector length*sin(angle).
I did not know you can do conditional operator-kind of operations in this part of Blender. Many thanks for this tutorial!
extreme complex but very interesting ! Thx for this deep, professionell Look in this "Industrie-chapter"
Great Job !! That is really excellent. I love your presentation. big thanks
Awesome content. Love that Blender is growing more and more as a tool in the engineering design and visualisation space. Getting an insight into the workflows and thought processes of high-end professionals like this is gold. BCon LA have definitely laid down the gauntlet for quality, for the Blender Conference in Amsterdam to meet in October.
Awesome job, Odi! I would also love to see how you would create a forged carbon fiber shader. Similar to the one found on BMW Vision M Next. Thank you and keep up the good work!
Great demo! And I love the normal visualization.
Thanks! Looks very cool. And loving the in-depth info!! Is that demo car shape available for download somewhere?
Question: When I apply a shader and then modify its settings like color or scale, it appears to save all these changes to the base material in the library so now every time I want to go back to that base material its now changed. How do I lock the library materials so they wont change?
Thanks