Timeline for Is there any way to get the face/vertex normals as it is rendered via python?
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 15 at 14:18 | comment | added | Digital_Utopia | If you take the normal map and the blend file from the other question, and set the color space of the normal map to non-color it will appear fine. But if you enable Magic UV and have it select flipped UVs in the UV editor you'll see that approximately half of the UV triangles are flipped. If you set the color space to sRGB (and drop specular to zero, and increase rougness to 1) Blender will then show the shading issues related to those flipped UVs which appear as they do in both the game and 3ds max. | |
| Jun 15 at 13:40 | comment | added | Markus von Broady | "Blender will happily ignore said flipped UVs so long as you set the normal map's color space to non-color" - based on my understanding of Blender, this is just not true. I thought this is specific to some engine, but never saw that in Blender. Could you prepare a simple geometry, preferably just 4 separated triangles, and show them in 4 states in a table: sRGB, non-color as columns, normal UV, flipped UV as rows? Then if you embed such .blend we can investigate this interaction. | |
| Jun 15 at 12:53 | comment | added | Digital_Utopia | But it's not quite as simple as UVs determined to be flipped, and to make matters worse, Blender will happily ignore said flipped UVs so long as you set the normal map's color space to non-color (which in turn has led to quite a few instructions to do so). Unfortunately it needs to be set to sRGB in order to see the result of said flipped UVs, as merely looking good in Blender doesn't accomplish much. | |
| Jun 15 at 12:46 | comment | added | Digital_Utopia | As far as the problem - I'm referring to the triangle elements in the UV editor, made up of 3 sets of UV coordinates. I appologize if I have the terminology wrong. The problem is, in certain engines, improperly oriented UV triangles will affect the final calculated normals, as the normal map will be calculated based on the UVs in question. effectively flipping the normal map for the corresponding mesh triangles, leading to a similar appearance as if you flipped the normals themselves. | |
| Jun 15 at 12:37 | comment | added | Digital_Utopia | My appoligies for coming off as rude, but that response was directed at someone who insisted I add the imgur album I linked in response to the comment you're referring to, in my question. I decided at the time that it really wasn't worth adding it to my question, for the reason I stated. I later changed my mind and added said album anyway... though my explanation for originally choosing not to seems to be validated. | |
| Jun 15 at 12:30 | comment | added | Markus von Broady | Though maybe the winding order is exactly what you mean. | |
| Jun 15 at 12:29 | comment | added | Markus von Broady | In your previous question I wrote "Honestly, I don't understand the question[...]" to which you responded "[...]I've already clearly described by problem and stated my question so that anyone capable of providing an answer can do so[...]". So this is your main problem, your attitude is severely filtering out people who can potentially help you. Next problem is, you use some terms like "UV face" and it's not clear what you mean by that. I think you never mentioned the engine you're exporting to? In Blender, normal direction depends on winding order of vertices, not on UVs. | |
| Jun 15 at 2:35 | comment | added | Digital_Utopia | You have to keep in mind that when normal maps are set to the non-color colorspace, Blender will happily ignore flipped UVs when calculating Normals. Unfortunately the target engine that this mesh will be rendered in is quite a bit more picky about it. For the purpose of the example, I learned that setting the color space to sRGB while dropping the specular and cranking the roughness results in a visually simular result to how it appears in the game and 3ds Max. | |
| Jun 15 at 1:43 | comment | added | Jakemoyo | I added an answer on your previous post. I don't think that this issue is caused by what you think. Kind of an XY Question IMO. | |
| Jun 15 at 0:34 | history | asked | Digital_Utopia | CC BY-SA 4.0 |