Intimate men (skin test)

[o4saken]

Testing a skin texture with skin pore bump map..

[wolgade]

Looks quite convincing. Did you use a procedural texture to create the bumps? Did you get away with the standard (2048x2048) resolution?

[o4saken]

Looks quite convincing. Did you use a procedural texture to create the bumps? Did you get away with the standard (2048x2048) resolution?

its just on the standard 2048, am quite happy with the results

[wolgade]

I would have guessed, that you needed more resolution for details. Nice to know. And how did you do it? Bake a procedural texture?

[o4saken]

I would have guessed, that you needed more resolution for details. Nice to know. And how did you do it? Bake a procedural texture?

honestly i was testing this to see what the female normal map looks like in the face ( it is the same one that is under more detailed female texture - WIP), and if it was worth pursuing putting it on the male normal maps..

Yes i do think that it would be better with a larger texture, but it works - how did i do it..
I took an image of white noise, then made it a lot smaller, tilled that in a 2048x2048 image. then applied this as a normal map in blender using the GLSL shading view - this allows you to see how it is looking in real time - so you can tone the strength of the normal map down till you think it is appropriate.
Now obviously the skin bump will be on the nails, under hands, ears - places you dont want it, so switch to texture paint - then smooth brush and go over all the parts that should not have the bumps.... once that is all complete and looking good i then baked it to the female normal.

One more thing i want too look at and am not sure if it is worth it - when in blender (unsure about other programms) and you have an object with say a 1024x1024 diffuse texture that this said object is uv mapped to, but you then change the texture to one that is 2048x2048 it automatically stretches the uv map to the larger size - which i have noticed works for multiple textures as well.. eg. if your now object is still using the 1024x1024 defuse texture, but you create a 2048x2048 normal map, it will map the 2048x2048 exactly the same as the 1024x1024... this means that although the skin texture is standard 2048x2048, i can work with a larger image for the normals details and it will map correctly.

[gaiha]

... if your now object is still using the 1024x1024 defuse texture, but you create a 2048x2048 normal map, it will map the 2048x2048 exactly the same as the 1024x1024... this means that although the skin texture is standard 2048x2048, i can work with a larger image for the normals details and it will map correctly.

Not a procedure I would do myself, as doubling the map size from 1024 => 2048 will exponentially consume a larger amount of your video cache, just to render that single image. Standard practices dictate image resolutions of 512 x 512 px in *.jpg format work best on MOST machines. if you consider the mathematics behind this...

using the Plane Area Formula of: h x w = A

512 x 512 => video cache benchmark and best practice.
h=512, w=512, then A= 262144 px
1024 x 1024 => has 4X the number of pixels than 512 x 512
h=1024, w=1024, then A=1048576 px
2048 x 2048 => has 16X the number of pixels than 512 x 512
h-2048 w=2048, the A=4194304
Therefore - your hardware has fewer resources available to render everything else in your model.

Subsequently, when you consider the 2D images you are using for diffuse and texture must now be mapped in 3D space...
Each pixel must now be plotted using the mathematical equation:

ax + by + cz = d
where each variable assigns the plotting point for each pixel

You've consumed 16x the processor time to calculate this, in addition to your video cache resources.
Just my opinion...

[o4saken]

Not a procedure I would do myself, as doubling the map size from 1024 => 2048 will exponentially consume a larger amount of your video cache, just to render that single image. Standard practices dictate image resolutions of 512 x 512 px in *.jpg format work best on MOST machines. if you consider the mathematics behind this...

Hmmm this is why i did say that it is something that i dont know if it is worth it... Not that i had known the reason i would have just tried and tested .
\

[wolgade]

512 x 512 => video cache benchmark and best practice.
h=512, w=512, then A= 262144 px
1024 x 1024 => has 4X the number of pixels than 512 x 512
h=1024, w=1024, then A=1048576 px
2048 x 2048 => has 16X the number of pixels than 512 x 512
h-2048 w=2048, the A=4194304
Therefore - your hardware has fewer resources available to render everything else in your model.

512x512 best practice? For what? Definitally not for face details of a MH character using the standard uv layout.

You only need these details for close-ups, but in this case your texture shouldn't have less resolution than your rendered image. Example: You want a rendered image with 2k resolution (1920x1080). With a 512x512 texture you have 140x140 for the face (standard uv layout). These 140x140 will be blury as hell in your rendered image. They're simply too low res to store the bumps o4saken is dealing with.