Skin Texture size difference between PS and SL

I am new at Creating Skin Textures.
I have read all the Posts in the Forum without finding anything about this Problem. I have completed several Tutorials on Skin Textureing and again found nothing on this problem. So its ether been totaly overlooked or so trivial that it has little importance to be mentioned. Then the answer could be so obvious and I am being totaly blind.
I create a sample of the skin texture that I wish to use and it looks great in Photoshop CS2, but once I load the texture and place it on my AV the texture is like it would look if I zoomed in on the texture 400%, createing a large blotchy looking skin texture.

I have tried to build my skin at 3000X3000 and at 1000dpi in PS. However I must reduce the texture to 512X512 before uploading into SL, and still I see the same thing.

Question 1: what can I do to adjust the texture cellure size, so it reflects that change in SL and show a clean (none Blured ) yet very small Cellure structure?

Question 2: I see areas on the AV where the texture streaches. ( ai..shoulders, back of arms, crotch area) What can I do about this?

Thank you in Advance
Shjak Monde

Skin texture blurring is caused by stretching and there's really nothing you can do about it but try to avoid putting much detail in those areas so it's not as obvious. The tops of the shoulders and insides of the thighs are the most problematic areas. Using a 512x512 texture means sacrificing some sharpness and detail but it's a necessary evil. If everyone was using 1024 skin textures it would negatively impact everyone around them.

As for relative scale of any base pattern you're using, you need to adjust it for each part of the templates. For example the back of the torso is smaller than the front on the template so you need to scale down the base pattern in that area to compensate. The head polygons receive more pixels of texture than anywhere else on the body so you need to scale up the base pattern on the head to compensate for the difference between the head and torso. And so on. It takes some experimenting to figure out the best relative scales to use, but once you do it gets easier.

A good way to do it is to create a checkerboard texture and apply it to all three tattoo slots. Take a screenshot from a fixed distance and compare the pixel width of the squares on the various parts of the body. From that you can derive percentage values to use.

Wow Thank you Chip that is some very important Information.
Now one of the Base questions was how do i regulate the texture sizes.. I mean what effects the size of the texture that will show up in SL?

such as:
In Many of the tutorials.. the author states they work on their texture at 3000X3000.. and then reduce it to 512X512. Does this reduce the size of the Cellure Texture Pattern when reduceing the size of the picture? If Not.. why don't they just zoom in?

Or..
Being I am useing the NagelSeries27 brushes for Photoshop CS2, to create my Textured Layers... Does the size of the brush more apt to effect the size of the Cellure Textured Patterns and show that change in SL?

I definatly agree with you that I have placed to much noise or texture in the skin for best look in SL and will try and work with a lot less. yet with what I do have I would like to stop it from bluring so much.
Useing your Guide in your last post could I like pinch the texture to such a point that once SL streaches it... it doesn't look streached?

Thank you so much for your help
Shjak Monde

When you size down the 'final' image from 3k to 512 pixels square, it basically reduces amount of detail (precision, sharpness) contained in the texure, rather than relative scale of patterns.

The image is always wrapped around the avatar in the same manner regardless its original 'physical' size. So, the difference is more like if you have one image that's 3k x 3k, one 512x512 pixels and one 256x256 pixels large, all produced by scaling the same source image... then the game client will stretch each of them to 512 x 512 pixels texture, apply optional layers of clothes avatar might be wearing on top of that... and then will wrap this picture around the 3d mesh. This means image larger than 512 x 512 will lose some detail, while 256 x 256 image will have each pixel of original image covering relatively large area, resulting in blurred look.

Joannah's answer for that was spot on.



You can put a lot of detail on most of the avatar without it looking too smudged or stretched, but there are a few problem areas where it's best to reduce it. I wouldn't reduce detail on the rest of the body. One thing you can do is just reduce the contrast in your noise patter in the areas of bad stretch but leave the overall tonality the same. The reduced detail is usually preferable to really obvious stretching.

In the really bad spots like the top of the shoulder and upper inside thigh you can try and pack more detail into those polygons to compensate for the stretch but it doesn't really work. Some of those polygons are only receiving 3 or 4 pixels of the texture map so you're better served by having those 3 or 4 pixels be the same.

Another benefit of applying a checkerboard (or the alternate clothing templates) to your av and examining it is that you can see exactly where the bad spots are. The fuzzier the pattern in any given area the more you want to avoid putting much detail there.

ahhh ok its starting to come a little clearer.



I see it would be best to simply work with the 512X512 and just zoom in if I need a closer look.
Fewer overall Cellure Textureing and little if any in those problem streaching Areas.

Ok thats what I will do then... I do have a problem of over doing everything... seems this is one of those times.

Thank you Chip and thank you Joannah
Shjak Monde

That's a good way, yes.... or you can work with a bit larger image (say 1024x1024) and then once you're done scale it down to 512x512 for upload... this way your final texture can have a bit more detail, that'll come from graphics program averaging groups of nearby pixels while scaling.... than you'd normally be able to put by hand into more 'coarse' pixels. ^^;

I created the checkerboard texture as Chip advised and ..!Yikes!.. there are some big differences in areas.
I took some pics and will use them as Guides.
I would love to post the pics for others to see but not sure how to upload them into the forum.

mmm Joannah?




Are you meaning that from 3k to 512 there is detail loss but from 1024 to 512 you gain detail?
Thats a bit confusing to me.

Ack, sorry ^^

the larger is original image you work with, the more detail can be put into it. So, 3k x 3k original will hold more detail than 1024 x 1024 original, and both will have more detail than 512 x 512 original.

Now, when you scale this image to the final 512 x 512 version, some of this detail from larger images gets lost. In case of 3k x 3k image, each area of 6 x 6 = 36 pixels becomes just single pixel, and each 2x2 (4 pixels area) of 1024 x 1024 image becomes one pixel. Because of this difference, small 1-2 pixels large details of the 3k image (like e.g. narrow lines) are very likely to get completely lost. On 1024 x 1024 image, the image resolution isn't really good enough for such details, but you still have 2x as much precision as you'd have working just with 512 x 512 pixels image... and even after scaling down quite a bit of these detail can be retained. So it's sort of compromise or middle ground between amount of detail you get to keep in final image, and the precision you can work with while well, working.

This is just an option though, not necessarily the best way to work... that i figure is up to each drawing person, separately ^^;;

Ahh ok ...so there is a point of definitive loss.
This is very important Information. Because I do want the very best possible.

Do you have any tips on the DPI?
At the Moment I am set at 1000 dots Per Inch.
Doesn't seem anyone else is doing that. Is there a reason?

There's no need to worry about DPI settings at all. Dots Per Inch is an instruction that's only meaningful when printing an image. A 3000x3000 pixels set at 300 dpi would produce a print that's 10 inches by 10 inches. At 1000 dpi it would produce a print that's 3 inches by 3 inches. The more pixels printed per inch, the sharper the print. There's a similar thing happening in SL. You can think of each section of the avatar (head, upper body, and lower body) as a fixed print size. The more pixels you pack into that space the sharper it will look. The reason we don't try and pack as many pixels in as possible is because the size of the textures that need to be downloaded by everyone who has you in their view distance expands exponentially the bigger dimensions you use. 512x512 seems to be the best compromise between performance and detail.

You only need to worry about the raw pixel dimensions of your textures. I usually work at 1024x1024 and scale down to 512x512 before uploading to SL. There are two reasons to work larger than your intended final product. First, it helps with matching details across texture seams. If you're working at 512 then your details need to match by the exact pixel. If you're working at 1024 you can be off by one and when you scale the image down it will still probably look like a correct match. The second reason is that it's just easier to work with more pixels. I think it's easier to get a more detailed look at 512 by working larger and scaling down. It's really a personal preference though. I find that scaling down from a larger image hides some of my flubs, and the ones it doesn't hide are only half as big then! hehe

Thank you much guys your the greatest