Often, when I'm texturing a model now in Blender, I'm confronted with the decision on what size to make the texture map. Now Blender defaults to 1020 p for their texture maps, but is that the most common size for a texture map. On the other hand, I've seen people swoon over aa texture map that is 4008 x 4008.

So what's the most common map size, and more importantly, what's the best.

Texture maps : The Commons :

256x256
512x512
1024x1024
2048x2048
4096x4096
=> these are common, and standard in most 3D apps

NOT Poser related limitation : Some apps can not go higher then 4000x4000

Poser (as from Poser 10/PoserPro 2014) can preview and use higher texture resolutions and 8192x8192 is the max used (my) preview window. (Depends on the installed video card and drivers)

Higher resolution textures can be used in the firefly renderer.

PS : my Blender defaults to 1024x1024.

1020x1020  is not a standard.

The best? ? ?
The bigger the texture size and resolution, the better the end result will be.

But stick to the standard sizes : 2 times 2, times 2, times 2, times 2, and so on.
That are the 3D standards..

Best quality for all Poser versions compatibility is 4096x4096

Quick Google search:

Standard Texture Sizes

Most graphics hardware requires that your texture images always be a size that is a power of two in each dimension. That means you can use any of the following choices for a texture size: 1, 2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, or so on (but unless you have a really high-end card, you'll probably need to stop there).

The textures don't usually have to be square: they don't have to have the same size in both dimensions. But each dimension does usually have to be a power of two. So 64 × 128 is all right, for instance, or 512 × 32, or 256 × 256. But you can't make a texture image that is 200 × 200 pixels, since 200 isn't a power of two.

We don't have the power of two requirement because Poser isn't a hardware renderer. Also, even if it was, a 200 image gets loaded into a 256 buffer and it still works. A 2000 goes into a 2048 and still works.

However, if you make an image 200, you consume the same resources as a 256, so why not make it 256 and get a better result.

The powers of two are still used for mipmapping the texture to different sizes on the fly. That is why internally they still get mapped into a power of 2, so they can be divided by 2 many times.

Of course you can divide 200 by 2 and get 100, then 50, then 25, then 12.5 whoops no such thing as half a pixel, trouble ensues, etc. The divide by 2 is simpler math if the software can start with a power of 2.

4096->2048->1024->512->256->128->64->32->16->8->4->2->1

On a side note, and BB also touched the small numbers :

The small numbers ? => Remenber the bucket size setting in the render settings?

A bucket is the square Poser renders per CPU tread.

On a 4 tread CPU, you will see 4 buckets rendering in harmony when you cick "render".
When a bucket is rendered and released to screen, you will see that tread starting the next bucket.

On a 4 tread Hyper treading CPU , you will see 8 buckets rendering together, and so on.

For bucket size use : 4, 8, 16, 32 (the default) , 64, 128, or 256 if you have a FAST CPU and Lots of RAM.

Bucket size also works most economically when you stick to the "power of 2" rule.

Hi

It all comes down to what you want to get out or your renders.

I just default to 4000 by 4000, or at least 4000 for the longest side of the texture for pretty much everything I do.

The higher the resolution the sharper the end result.

That's not to say that it isn't fine to use lower quality textures on things that are in the background to save overhead.

But if the final render is going to be UHD then it's worth making the textures super sharp.

Once you save the master in a lower resolution it's gone forever, so I always keep my masters saved at their full resolution.

Mike

Pros only use 1024x1024,2048x2048,4096x4096,8192x8192 cause it's the industry standard.

U.S. Feet & Inches .EURO DM & CM

for a Poser dress I would make a 4096x4096

& maybe a 2048x2048 for the hardware challenged users.

but when ya reduse all the 4096x4096 textures to a 2048x2048 I would check the seems.

Interesting (or not) twist - I was trying to retexture a swimsuit (with stripes) in PP14, and the textures was always badly smeared. First I copied the vendor default texture size of 2500x2500, then I reduced to 2048 and that did make a good difference. Still not perfect, but clearly better at 2048x2048 than 2500x2500.

If you were using the old-style "Quality" texture filtering, bigger is bad. It was too aggressive and was doing extra blur when given bigger textures. Or, more specifically, each size reduction in the mip-map stack was worse as the stack was taller. By going over 2048, you added a layer to the stack.

back to the original question. The pro rule is that the visible part of the texture map (which might be half of it, or even less, as the other part is turned away fromt the camera) must deliver at least twice (*) as much pixels as the space it takes in the render.

Example: you make a 2000x2000 render of a face which takes half of the render, the face takes one quarter of the head texture, then that head texture should be at least 4000 (okay, 4096) because 1/4 x texture 4000 = 1/2 x render 2000. But when only 1/8 of the texture appears in 3/4 of your 2000 pixel render, you already need a 6000 wide texture.

So when your object is in the distance only, you don't need the large textures. And when you go large scale fine print of close-ups, your textures will have to go BIG.

(*) this relates to your Poser shading rate, as set in Render settings. "at least twice" implies that your shading rate is < 1/(2x2), and that each renderpixel can be 2x2 supersampled for anti-aliasing so pixel samples =2. Setting your pixel samples to 3 (3x3 supersampling) implies shading rate about 1/(3x3) so 0.1 will so, and requires a texture which it at least 3 times denser that your render so the numbers dicussed above go up by a factor 3/2=1.5

So texture size, render size, AntiAliasing (Poser Pixel samples) and subsampling (Poser Min Shading Rate) are sort of interrelated. You can pick any value you want for each, but when you want best quality at best efficiency, each of them puts limits on the others.