Optimal texel size VS Texture Utilization in the tutorial

Development Asset Creation
,
1

Hi, In your tutorial below,

https://youtu.be/wA3uqRYY2NY?t=595

This guy is referring to the Unwrap result as “It Cannot be better”  when the texel size matches the Optimal texture size, but the texture utilization is only “28%”.

This sounds odd, because my understanding is that texture utilization is quite critical for the overall quality of a texture, and in most cases, higher texture utilization is recommended to get better results, as explained below.

https://youtu.be/2-mIY87314g?t=96

In Reality Capture, is “Optimal texel size” more important than “texture utilization” ? If so, why is this? ( Is “texture utilization” in RC something different from the commonly known “texel density” ?)

 

 

2

Hi YUJI,

the mentioned “can not be better” was related to the texture quality. It is important in this case. Yes, there is also texture utilization, which can be achieve with a  different settings, but the quality is more important.  

Also, the utilization ios so slow because of the wanted resolution. It will be higher wit smaller texture resolution. 

3

Hi ,

Thanks for helping me. I think I understood, but could you answer 3 more questions regarding the unwrap result in the video?

Please look at the result of the 2nd and the 4th try shown below.

[2nd try]
[)

Texture resolution: 16K
Count of textures: 1
Texture quality : 157%
Texture Utilization: 67%
Unwrapping style: Maximal texture count
Optimal texel size:0.000273 units per texel
Texel size:0.000173 units per texel

[4th try]
[)

Texture resolution: 16K
Count of textures: 1
Texture quality : 100%
Texture Utilization: 28%
Unwrapping style: Fixed texel size 
Optimal texel size:0.000273 units per texel
Texel size:0.000273 units per texel

Question 1.
If “the quality is more important” as you say, the second result (157%) seems better than the 4th result (100%), but it’s obviously not good right? What’s wrong with a 157% quality texture? Does it lead you to bad results, like corrupted textures or large texture size or something?

Question 2.
If the “Fixed texel size” option always gives you the optical texel size, and give you 100% texture quality, like in the 4th try, why don’t everyone just choose “Fixed texel size” instead of choosing the “Maximal texture count” option? Could you tell me the pros and cons of these 2 Unwrapping styles?

Question 3.
I understood that the “texture quality” is more important than “texture utilization”, but could you tell me any cases that higher “texture utilization” becomes important?

Or should I just care about the “texture quality” and don’t have to care about the “texture utilization” at all?

4

Why do you think, that 157% texture quality is wrong? I know, that 100% should be maximal, but in this case it is not. According to number, 157 is higher than 100, but 100 is also enough.

The option for texturing is on you personally. But we recommend to use Fixed texel size to achieve the best possible textures. In Maximal texture count the texel detail is adjusted automatically so that the texturing fits into the selected maximal number of textures within the selected maximal resolution. If you want to control the texel size use the fixed or adaptive texel size style if your intention is to deliver a model with pre-defined visual precision, e.g. 1cm resolution for creating a true ortho-photo map. There will be so many textures how many are needed to texture the model with the selected texel size. 

For practical reasons, e.g. if you are scanning a statue on a wider square, you may want to have finer details closer to the statue and 1cm detail elsewhere. Choose the Adaptive texel size style where you can define ranges where your texture must fit. The RealityCapture unwrap algorithm will use the 3D model, camera poses and inputs resolution to calculate optimal texel sizes and it will clamp the estimated texel into a range defined by:

  • Minimal required texel size The smallest dimension of a texel. This is useful when moving with a high-resolution camera between objects to avoid wasting texture pixels.
  • Maximal required texel size Define a maximal texel size, i.e. there will not be bigger texels. This is practical if you want to paint into the texture later.

I think it could be in game industry or if you want to render the image. But it is unnecessary to use high texture with small utilization, In that case just use smaller resolution and the utilization will be higher. When you will receive this small utilization, then you have unnecessary high resolution.