Previewing avatar textures on the main grid for L$0

The main grid of SL is excellent for previewing avatar textures. You get to test your textures in their final environment under multiple lighting conditions. Most external previewers fall short at reproducing the ever-changing SL environment accurately, so this is an important tool in obtaining texture accuracy you may need for producing top quality skins or clothing. You may say “I’m short on cash, and can’t afford to do that”, but this preview method is simple and quite affordable (L$0) with the help of Photoshop and a little knowledge of the SL client “character” folder contents. THIS IS NOT THE PREVIEW GRID OR THE SL UPLOAD PREVIEW TOOL. This is a preview hack that allows full size avatar textures to be loaded onto the live server directly from the SL client.

The ideal preview workspace is a dual monitor setup with Photoshop and SL running side by side.



If this is not possible then stacking or splitting the windows on a single monitor is almost as fast. You WILL need a workstation that can handle the memory load! There is no getting around that! Running a preview is as simple as clicking on the Photoshop action “Play” button, then refreshing (rebaking) your SL avatar textures using the <client><rebake textures> menu button from the SL window. Make sure your avatar is wearing the DEFAULT LL SKIN while previewing. The client avatar color layers are stacked below any avatar textures you may have applied from your inventory, so they will be blocked by any non-standard skin textures and clothing layers that sit above them.

Before previewing can be done efficiently an action set must be created in Photoshop. Photoshop has macros built into its functionality in the form of “Action Sets”. Action sets consist of one or more series of “Actions” that can be saved to quickly perform complex or repetitive tasks that take a lot of time to do by hand. Previewing avatar textures inside SL is a very repetitive task, and a prime candidate for a set of Photoshop actions. A Photoshop action is set up just like a recording. In the Actions tab in Photoshop you will see the typical record, play, and stop buttons and symbols at the bottom. These are the buttons that are used to create and play an action. The preview actions that are required for previewing on the main grid in SL are extremely simple to set up. They consist of a set of actions that save each of the three avatar body textures to the correct location in the SL client “character” folder. For Mac users this folder is embedded in the packets associated with the SL client. To create the actions –


1. Open your avatar texture in Photoshop and click the <Record> button on the actions tab.

2. From the Photoshop drop down menu <Select All><Copy Merged><File New><Paste> (if your texture is oversized you will want to <Resize> it to 512x512 here)<Save As> and locate the correct file in the client “character” folder. The file will either be “head_color.tga”, “upperbody_color.tga”, or “lowerbody_color.tga”. Overwrite the current file with your texture for the corresponding body area (i.e. a head texture saves to “head_color.tga”). Delete the temporary file (the resized image) and <Deselect All> in your master avatar texture window. Once this is complete click on <Stop> on the Actions tab.

3. Your first new action is now complete. Repeat this process for the remaining body sections (upper and lower body) until you have a complete set of three actions.

4. You are now ready to save these actions permanently as an “SL Preview” Action Set. The command for saving action sets can be found under the drop down menu in the Actions tab. The next time you need to preview a set of avatar textures you simply load the “SL Preview” actions, if they are not already loaded into your Photoshop Actions tab, and play the correct action for your body area.

You can also set up a single action to save all three body areas at once if you have those textures stacked within a single texture file (see the above graphic). This action and file setup is a little more complex, but saves even more preview time.

Not all file configurations and personal preferences will be the same because of different software releases and different best practices, so not all actions will be the same. I have several action sets created for specific files and uses. The ones I use most frequently are the most complex because I deal with composite files that hold the head, upper body, and lower body sections in a single oversize file. Below, is a breakdown of a simple action I use for an oversized (2048x204 32 bit head file as it appears expanded within Photoshop's Actions tab. Some of the information can be ignored like Width, Height, DPI, and Compression because it is either default or has no bearing on the final output.



One very important consideration to take before altering the SL client “character” folder is to have a backup of the files located there. You can do this yourself by copying all the contents into a new “backup” folder, or use this handy little executable I’ve created here (http://www.secondskinlabs.com/Forums/SSA_Uninstall.exe) to restore all the original “character” files. If you lose the files you will need to do a COMPLETE REINSTALL of the SL client to restore that folder to its original state.

::looks enviously at your setup::

I notice like any seasoned computer tweaker the side of your case is off =)

I'm curious as a related topic, are the shading/alpha/wrinkle/etc guides in the character folder viewable by others inworld after a rebake?

YOU WACOM IS HUGEEEEEEE. Mines like so small. T.T

Your like my hero. You answered the last question I posted nicely, good information, and it seems like you KNOW what your doing. I respect that.
My workstation is like the poor version of yours. My old desktop, my new laptop, and my small wacom tablet surrounded by paper work, screw drivers, books, and other junk. lol

MAH HERO!

Yes, those files get updated every time the sliders are adjusted in appearance. Most of those *.tga files can't be altered without breaking the client (If that happens, it's best to set the character folder back to default).

It's the mid size 6x11 (the only one with a wide screen aspect ratio). It also fits nicely in a briefcase with my laptop, which is another reason why I got it.

Namssor, you forgot to add a very important step. I added this to the other thread, but I'll repeat it here since its the topic.

To see the previewed textures, you need to be wearing a clear skin. Just create a new skin and leave the textures blank, then wear that before rebaking.

Also, you don't need to paste merged layers into a new doc. If you have "save a copy" checked, PS will save just a flattened Targa.

You should also add an UNDO step at the end of your action so that you're back to where you started in one click.

It's also not totally necessary to downsize to 512 x 512. Ive been able to save 1024 px textures and see them in their full resolution inworld. This helps alot with finicky areas of the avatar as well as being way faster to run the action.

I also want to add that this is not just a client-side effect. Everybody else inworld will also see your preview skin in all its unfinished glory.


Namssor, I do want to thank you for teaching me this awesome trick in the first place. Its been one of the most useful tips i've learned on the forums by far, and its made doing makeups much much easier for me.

True. It was major assumption on my part to skip over that detail thinking it was obvious. I'll be going back and refining the step by step to include more of the details (of which quite a few are missing).

I'm aware of that. I think this is a matter of preference, and I'm not going to discourage anyone from using the save "As a Copy" option. I just don't like it. It saves a similarly named copy to disk, and I'd rather keep the entire opration restricted to the clipboard away from the original and away from the files on disk until absolutely needed. Also, if you intend to cut sections out of stacked texture files, this is a best practice. Resizing and cropping an original is something I regard as a risky operation. Therefore I opt to do that on a copy I create that has not been saved to disk anywhere yet. It makes things easier for cleanup too.

Yes, I forgot to mention some of those details. What I do is delete/close the copy window and "Deselect All" in the original at the end of each action. The action runs and leaves no trace upon completion.

Unless something has dramtically changed recently, I have never been able to see full resolution 1024x1024 textures on an avatar. For me the difference is negligable since the textures get served up at 512x512 regardless of whether the client side textures remain at 1024x1024 or not. The difference you may be seeing is the fact that the TGA files do not undergo a lossy compression before getting served up to you between rebake and that 30 second redraw period, unlike all the textures that get uploaded the normal way (through the client menu) to the server first. For that reason they are slightly better quality than the ones that eventually get redrawn after a bit of lag time. The differences though are nearly imperceptible. 1024x1024 textures are also going to eat up resources unless they are resized from the start. I work at 4096x4096, which is not an option. I must cut and resize my textures.

Sorry again for this omission. It was somewhat intentional because it's not really neccessary to know all the facts about this in order to preview the textures yourself. Once others start getting involved, yes this aspect of previewing becomes important; However, it needs a bit of further explanation. It is not as intuitive as it sounds.

There is a different lag time between what you (the one holding the updated textures on your client) will see and what others will see, and when. For you the update appears almost instantaniuosly, while others may need to wait up to 30 seconds for the compressed redraw to happen. You can monitor this transition point because the avatar textures will suddenly go blurry and slowly redraw for you as well. This takes place at various speeds depending upon a whole bunch of factors (client system settings and capability, connection speeds, etc.). What is fundamentally important is that IT WORKS.

The client redraw transition brings up another important point. It is possible to break the texture rebaking function by hitting it too frequently before the texture redraw. If this happens the best way to fix this is to go into preferences and toggle the anisotropic filtering on or off depending upon what it is set to. This triggers a reload of the entire scene rather than just the avatar textures. If this fails (and it has for me on occasion) the next step is a relog.

You're quite welcome! This whole experiment started close to 4 years ago for me, and I can honestly say I'd have probably left long ago out of frustration were it not possible to freely and quickly preview my textures in world as they are meant to be seen.

We definitely have different workflows. I'm not really too afraid of killing my original file, since I version it every session, and the action saves to different directory than my source file, so it won't overwrite the original. Plus I really like having a single button that will save off a preview and return me right to where I was in a few seconds.



I don't know where the resizing is happening, but I definitely noticed a very big difference in quality. I just noticed this last week when I was working on eye details with the template grid layer visible. I work at 1024 resolution on my main files and when I previewed without resizing, the gridlines were very much sharper than when I resize to 512. It could have been just a difference in compression, but it definitely looked like higher resolution. I'll try and get screenshots for comparison next time I try it. I would only use this sparingly for very tricky areas where extra resolution is needed. And also, I don't think its possible at higher than 1024. There wasnt any difference when I tried it at 2048, and I don't fully remember, but it might not have rebaked at all in fact.




True, but its also very handy too. I sometimes like to work on my skins while sitting around with some friends in one of our workshops, and I can get feedback from them as I work, which is a huge huge help. (It also means they see an awful lot of me looking very strange with gridlines and only one eyebrow or half my face made up, but they're used to that by now.)





Thanks for that tip I had this happening to me the other day. I found if I changed through a bunch of other skins, I could finally get it to rebake again. I'll try your method next time.





You're the best!

I'd be very interested in seeing some comparison screen shots if you get some. What type of interpolation are you using for the image reduction when you resize it to 512x512?

I may be missing a lot here and I glanced over what most of you were talking about. But wouldn't resizing from a larger resolution to a smaller resolution cause blurring and take out any details you may have done at a higher res?

So why not just start on a 512X512 image and save yourself the hassle. Unless your counting on blurring to occur in order to cover up any mistakes. >_<

Don't start work on a 512x512 image as a master file. This is the wrong approach. Detail is much easier to manipulate at a higher resolution and much easier to maintain when reduced through the proper algorithms. The people that illustrate postage stamps don't paint actual size for the same reason. It's just too difficult to manage details at such a small size.

There are good ways and bad ways to resize an image and some blurring is actually your friend. It can add more detail or fool the eye into seeing more detail where there is none.

Antialiasing is a perfect example of good blurring. Ever wonder why some screen text looks better (less jagged) than other screen text? Antialiasing smooths (blurs) the edges to fool the eye into seeing a more naturally defined edge on screen text. This can be applied to avatar texturing too. When I reduce freckles on a skin to a mere pixel, they don't actually get reduced to a single pixel. They get reduced to nine pixels with eight of those surrounding pixels averaging the values to make the freckle appear rounder rather than square. This makes a difference!

Another example of good blurring is when we enlarge an image with bicubic interpolation. Bicubic interpolation averages the values of surrounding lower resolution pixels to create a smooth transition between higher resolution pixel data. Bicubic interpolation is a BAD THING for image reduction. It's like trying to file against the grain, and actually hurts the image quality during reduction. It would be like adding sixteen pixels to render a freckle when it will look much sharper with nine. Nearest neighbor interpolation is the proper way to reduce an image. It will not blur the image. When an image is reduced with a nearest neighbor average there is obviously a loss of detail in the number of pixels, but not a loss of detail in the way we see those pixels next to one another.

The best way to texture an avatar is to work at high resolution and reduce the resolution before uploading or previewing with nearest neighbor interpolation. 1024x1024 resolution should be the absolute minimum image resolution for a master avatar texture file to take advantage of good detail reduction. I highly recommend 2048x2048. The level of detail control improves even more. Think about it - 1024x1024 buys you 4 pixels for every one you end up with on the final texture. 2048x2048 buys you 16 pixels!

Boy, Thats going to eat my computer alive at those resolutions. LOL.
I was going to also add that maybe leaving it at higher resolutions may be better but that would cause greying and missing images. Or it will take to long for a skin to load on a character.

I think you're misunderstanding me. The preview textures are sized down to 512x512 off of the master file, which is either 1024x1024 or larger. Anything that gets uploaded to SL's servers is no larger than the default 512x512.

Now, if your computer can't handle 1024x1024 images displayed on it's own monitor then you have an entirely different problem. But, since I know you run SL on your computer I know you don't have that issue.

The size of your working file is just a matter of personal preference. For me, anything larger than 1024 x 1024 would bring my computer to its knees. ( My head file is up to about 150 layers and is over 250MB. ) Plus, I know if I were to work larger, I would get too carried away fussing over little details that wouldn't ever survive resizing. 1024x1024 keeps me focused on the texture as a whole, while still giving me plenty of detail.



Namssor, when I read this, I almost spit my coffee out. In 15 years of using Photoshop in professional print environments, I have NEVER heard of anyone NOT using bicubic interpolation for downsizing images.

But then I tried Nearest Neighbor interpolation just out of curiosity. And you're absolutely right. The details hold up much better. I still would use bicubic on photography, but for hand painted images like skin files, I'm now a nearest neighbor convert. I love how no matter how long I use PS, I can still learn new ways of doing things!


p.s., I've been busy getting my new store open, but next week I should be able to post some screenshots of the skin textures previewed in-game at 1024.

There's nothing wrong with that.


The newer PS CS2 and CS3 releases introduced a newer version of bicubic interpolation that handles image downsizing better, so I should probably mention that so I am not misleading people. "Bicubic With Sharpen" interplation was introduced, and this is also good for image downsizing. In fact, it's recommended. Still, if you are using an older version of PS without those options nearest neighbor will work better than the standard bicubic interpolation.

Should you save as TGA 1024x1024 first then resize to 512x512 bicubic sharpen? Or should you resize in the original PSD file then save to TGA?

It doesn't matter. TGA and PSD files are not compressed. The only reason to save to another file before resizing is to protect your master file from accidently being overwritten. Using an action set to automate the process eliminates that possibility.

Export to TGA first, then downsize. There are a few reasons for this.

First, it will ensure that things like layer effects do not get wonky as you downsize. For example, let's say you've got a 5-pixel stroke around something. When you downsize the image, the whole thing will get smaller, but the stroke will remain 5 pixels thick, since that's its setting. You'd have to go back and manually change it to make it the right size. But you couldn't even do that either, since the number 5 is not neatly divisible. You can't have have a thickness of 2.5 pixels, or 1.25 pixels, of course.

That kind of thing can't happen with a TGA. Since it's flat, there's no such thing as layer effects. Everything will have been rasterized and merged. When you downsize the TGA, everything, including that stroke (which is no longer actually a stroke) will downsize uniformly. There won't be any need to adjust any effects parameters because there are no effects, and no parameters. It's all just one raster. And the problem of the fractional pixels will be solved (at least in appearance) by anti-aliasing.

Second, it takes a hell of a lot less memory and processing power to downsize a flat image than a layered one.

Third, there's less margin for error. If you downsize your working document, and then save it that way by mistake and exit, there's no going back. All your layers are downsized; the information for their larger selves is gone. That's it. But if you never resize your PSD's you never take that risk. Even if you totally screw up a TGA by resizing it incorrectly or something, who cares? You can always export another one if the PSD is still good.

All good points, Chosen. I kind of assumed that Snick was flatening the file before resize, and I should have made that clear in my reply just in case she wasn't. Thanks for clarifying.

First of all, I love this technique. I've used it for awhile now & it's saved me so much time and helped me so much. So thank you for this post, where I originally learned this technique, which I'm now bumping because I have a question...

I'm on a Mac (still on Tiger), using CS2. On a Mac, the files you need to replace are located within the .app package file. Which you can navigate to via special command in finder, but you can't get there through the standard file chooser menus in photoshop. And I can't figure out how to hack the file location in my photoshop action to point to the proper (actual) character folder. So when I use this technique I export to another regular folder, and have to do an extra step with each export where I must copy & replace the file(s) from the folder I do export to, to an open finder window of the proper folder (and hitting the "yes I want to replace this file" button), then switch to SL and rebake. I know it doesn't sound that bad but after awhile that extra silly step really bugs me and slows me down.

So I thought I'd ask on here if any Mac users who use this technique have any suggestions/help on this one

Two tricks that I've used for down-sizing and sharpening.

The first one is to copy and High Pass. Then down-size and use the HP layer to sharpen where I feel the need (Layer Mask). When doing this trick, I HP and then down-size. I muchly prefer this order-of-operation.

The other one is View > Arrange > New Window for...

With a new window open, you can have one copy of the window for mad zooming and detail painting while keeping an eye on it at the final size. For example, if your working document is 1024 and your final will be 512, then zoom the new window 50% for a damn good approximation of what it will look like when you are ready to down-size and export. Awesome tip if you have the screen space and are comfortable with detail pixel painting.

I'm CS2 and I believe it was View > New View in older versions. Don't know if they changed this for CS3 or CS4. Shouldn't be too hard to find.

I don't know if this is widely known or not but in the current version of the Greenlife Emerald Viewer there is an option to upload a temporary texture to the main grid. When you upload your texture there are two check boxes at the bottom of the preview window, one of which is for a temporary texture.

When checked off it uploads a single texture at no charge to the main grid and puts it in your inventory. The texture appears to be available for the duration and is shown in italicized text in your inventory with the word temporary beside it. I'm not sure if there is a maximum but I have uploaded entire outfits using this and previewed them live on the grid at no cost.

Additionally they are visible to all other Avatars so others can give feedback on them if you wish.

for those hesitant to try out a viewer not from Linden I would suggest giving this one a go. It has a number of very VERY nice features in addition to this one and they keep up with the updates being applied to the Linden official viewer.

BTW - I am not affiliated in any way with these folks, just your average dumb user with a photoshop license.

Thanks for the heads up, Dante. Add this to the (long) list of features the LL viewer should have had for ages now. Hooray for open source.

My guess is it's not actually upoading anything until and unless you pay the L$10. It's probably just putting a temporary copy in your local inventory cache, as well as locally rebaking your avatar outfit. And since the baked outfit is distributed to the sim, other people can see it. Theoretically, one could do this forever, and never actually have to upload clothing at all (at least until LL finds a way to plug the loophole).

You could actually do this same thing by hand, as well, simply by replacing the files discussed earlier in this thread with your own composites. It wouldn't include any components for which you don't own the source imagery, but if you make all your own stuff it'll work just fine.

Bravo to Greenlife! It's about time! This is even better for those who work with the clothing assets, as they will get to preview the fabric morphs too. Also, those who work with prims have a preview option. Can't do that with my preview hack.

I have yet to integrate this client into my previewing process (I doubt it will be faster for skin previews), but it looks like another good tool to have at my disposal for working with prims and clothing.

Chosen, I was wondering how they were doing it, that sounds like a likely way to do it, I was wondering. It indicates that its uploading and when complete unlike the paid for option doesn't give you a preview however it does appear in your inventory and you can double click on it and display the texture same as any other in your inventory. I haven't tried to apply it to in world objects yet to see whether they would be visible that way. I guess those will be the next tests.

Nam, it is a definite boon for clothing makers! I went searching everywhere for more info and just stumbled across it by accident. The viewer itself is very stable and seems to be quite popular but this feature alone makes it worthwhile for designers. I've so far used it for skins and clothes. More tests to come.