uploading problem

hi
when i upload graphics they distort to 256 by 512. I read in the knowledge base that the system should be able to handle up 1024 by 1024.

Last night I tried to up load a pic that is 400 pixels wide by 390 pixels high and 150 dpi. It distorted to 256 by 512 and looks terrible.

Is there a setting that needs to be adjusted or something else that I am missing? Do I need to adjust the pic before uploading?

This should help Cleowolf!

/109/e6/150360/1.html

Try to adjust your texture/picture before you upload it into SL..Second life saves textures in 3 different height/widths that I can see..512..256..and 1024....If your texture say in real life is close to 1024 by 800...guess what..that 800 integer will be rounded down to 512 and your texture will come out distorted. SL doesn't crop the image..it stretches it or squooshes it. If you want to upload a large texture..try to make it close to the 1024x1024 size and let SL do the rest..Goodluck.

~Lana Tomba

lana thank you for the simple understandable answer.

so you are basically saying that sl will distort any pic that is not 256 by 512 or 1024 by 1024. every thing i upload will have to be adjusted to fit those sizes before hand.

that is disappointing. guess i wont be using much of my artwork in SL.

kim i read the page you linked to, wow what a byzantine system.

see if your original file is say 512 by 256..then no distortion...sometimes i take snapshots in SL and then clean them up and i use a widescreen monitor..my snapshots are always an original file size of 1680 by 1000. if I upload the original file with no cropping..the 1680 integer has to be "squooshed" or distorted..like approx 700?..pixels?...thats alot of distortion and the 1000 integer only has to be stretched like 24..so..i'll take the shot and then crop it trying to make both height and width as close to "even" as I can..then enlarge the image to 1024X1024. If your original artwork is larger than that..don't worry about its original size...only its dimensions.Ive been able to upload some 3k x 3 k pics but they're ultimately downsized by SL to 1024 X 1024.

You can always compensate for distortion on a prim by horizontal and vertical offset tabs combined with the two top tabs that stretch the texture out veritcally and/or horizontally(in the edit window of the prim "texture"..but ultimately you will lose some of your texture(what is actually displayed on the prim)..unless(and i do this alot) place the full texture on several prims and stretch it or squoosh it to correct the dimensions and sell the artwork on several prims linked rather than one.Say split the art in to four prims..offset the far left prim by -0.250..the second prim at 0.000..ther third prim by 0.250 and the 4th prim by 0.500 vertically then make sure you stretch the texture vertically uptop in 0.250 incraments as well.

~Lana Tomba

P.S. I've bought some artwork in SL..it was sold as limited numbered editions..the first piece was always the cheapest(because the first person to buy the art was offered a good price) and then the price increased as the numbers numbers got higher on the way to 6.(Naturally the 6 th edition piece was the most exspensive because the artwork had been distributed and was obviously "popular"..therefore the 6th person to come in and buy had to pay more.)Also this afforded the client to be able to buy all six pieces to make sure they had the "only" prints and copies..

There are a lot more available sizes than that. Any power of two between 8 and 1024 will work. Anything that is not a power of two will get downsized to the nearest power of two below. For example, 320x240 would end up as 256x128. 150x50 would become 128x32. Etc., etc., etc.

I would encourage you to read the guide that Kim linked. Every possible texture size is listed.


It will never stretch an image. The upload routine cannot upsize an image, only downsize.



Don't make it "close to" anything. Use exact powers of two, always. If you want your texture to be 1024x1024, size it to that in Photoshop BEFORE you even think about uploading it. Do NOT rely on the upload routine to resize your stuff for you. It won't do nearly as good a job of it (in terms of visual quality) as Photoshop will.

That said, you should almost never use 1024x1024's. The single biggest reason SL runs as slowly as it does is because people are irresponsible about texture size management. Consider that every 1024x1024 uses 3MB or 4MB (depending on bit depth) of texture memory. It doesn't take all that many of those on screen before they start to overwhelm a 128 or 256 MB video card. Always keep your textures as small as possible. Generally speaking, 256x256 is as big as you ever really need to go. Only go bigger if your texture has small text in it or other very fine details that would become illegible at 256.

For more information on responsible use of texture sizes, again see the guide Kim linked. The URL again is /109/e6/150360/1.html


Unfortunately "simple and understandable" does not always mean correct. (No offense, Lana.) Most of Lana's answer was incorrect.

Once again, acceptable texture dimensions are all powers of two between 8 and 1024. That means 8, 16, 32, 64, 128, 256, 512, and 1024 all work, not just the three sizes Lana mentioned. It's only arbitrary non-power-of-two numbers that get resized.


Hopefully, this better information is less "disappointing" for you. In any case, don't give up so easily. Take a look around. Thousands upon thousands of people create fantastic artwork for SL every day. There's no reason you can't be one of them. So you might need to resize a few things; so what? It's really not a big deal.


There's nothing "Byzantine" about it. It's just how 3D graphics works. It's been this way for many, many years, and it's got nothing to do with SL itself. The power of two thing is a requirement of OpenGL. The cap of 1024x1024 exists because some graphics cards are not equipped to deal with textures that are any larger than that. For a while, LL had upped the limit to 2048x2048, but they quickly moved it back down to 1024x1024 when it became apparent that some users' systems would crash whenever they'd encounter those huge textures.

Consider that the most common size for all video game textures is 256x256. I'm sure I don't have to tell you that the texture art in a lot of video games looks absolutely wondrous. There's no reason you can't do what those game artists are doing. The only difference between them and you is they've had more practice.

Remember, it's not things like file formats or canvas sizes that make or break the quality of your work. The only things that matter are your artistic ability and your willingness to understand and accept the rules of the medium you've chosen to work with. If you were an oil painter or something, you wouldn't complain about how the paint works or worry about the size of the canvas in front of you; you'd just use the medium however it happens to work, and you'd fill the canvas, whatever its size, to create your art, all without even really thinking about it. Do the same here. The available canvas sizes are what they are. Instead of complaining about them, just use them. You'll do fine.

sorry oh guru of the texture world ..I clearly stated from what I could see. Not once have i "ever" seen a texture uploaded in any other dimension other than 256...512 and 1024.
If they are..I've not seen it.I'm sure theres a legitimate reason why.

As far as making it as close to 1024 x 1024 as i could?..well it worked for meno error messages..no slaps on the hand about downloading in integers by two.

It's good to know the things you posted chosen.Thankyou..however conjecture is just that..conjecture..I was merely trying to help Cleo rather than bombard her with an 8 hour tutorial or 16 page tutorial on "why her images were distorted".

I did not know that images could NOT be stretched..strange that I've uploaded pics at 1000 and sl showed them to me as 1024...don't know how that happened either.But as I stated this is from my experience "or lack thereof" with textures.

Please..lets not try to teach the birds the nuiances of flight in the fall using the jet stream as a back wind heading south...when they dont even know how to fly yet.

My info was based on experience and that alone..."from what i can see".


sigh..i saw this coming... ..I'll leave now.

~Lana Tomba

hi Chosen

thanks for the help

i am an oil painter, blacksmith and leather craftsman and a few other random crafts. suddenly i feel like a scribe talking to gutenberg.

i had scanned a pencil drawing, reduce it to 400 pixels wide by 390 pixel tall (keeping the original dimension ratio). this was going to be a gift as a stand alone graphic.

on a side note i had been thinking of using some of my preexisting paintings and drawings to make in game posters and the equivalent of framed pictures for people to hang in their houses. basically scan, upload, frame and sell. would that be best discussed on the building forum? the reason for asking the question here is because this forum seemed closest in subject.

so basically you are saying the width and hight of the image must be any of 8, 16, 32, 64, 128, 256, 512, and 1024 and nothing else.

for textures keep it below 256 the lower with out losing unnecessary detail the better and try to use PNG file formats. only use above 256 is when you require high levels of detail.

once again thank you

lana i am a he or at least that is what my friends tell me

i stole my name from a seventh century saxon, i had no idea what to expect in 2nd life when i picked it.

Lana, don't leave. No offense was intended. I just always strive to do what I can to prevent incorrect information from floating around. If you're uncomfortable with having been corrected, I'm sorry, but really, there's nothing wrong with it. I get corrected too whenever I inadvertently say anything that's not factually correct, and that's great.

And this happens to be just such an occasion! You were right, and I was wrong about the upscaling thing. I just uploaded a 1000x1000 test image, and it did in fact get resized to 1024x1024, just as you said it would. I'm positive that that behavior was not always the the way it worked, as similar tests I've run in the past have all resulted in downsizing, not upsizing. I have no idea when or why they changed it, but it seems it was indeed changed nonetheless. Thanks for pointing it out. My apologies if I confused anyone.


Here's what I'd suggest. Re-scan it with a width of 512, or if the original scan is larger than that, down-sample it to that width. The height will end up at 499. Since 499's not gonna work, and since you don't want SL messing with the proportions of your artwork, "pad" the image by increasing the canvas size (not the image size) to 512x512. You'll end up with an extra strip of blank canvas space (6x512) at the top, and another at the bottom.

When you apply the texture to a surface in SL, set the vertical repeats per face to 0.97 instead of 1. The lowered repeat will stretch the image vertically on the surface, effectively trimming off the extra pixels. You end up wasting a small amount of texture memory this way, but it's only 12x512 pizels' worth, which is less than 20 kilobytes. That's not enough to make any noticeable dent in performance, even if you were to do it a thousand times.


Correct.

Technically, it CAN be something else, but if it is, SL will end up resizing it on you. Since SL won't do as good a job of the resizing as Photoshop would, it's best to pre-size it yourself before uploading.

Also, you probably realize this, but textures don't have to be square. 32x1024 is just as acceptable as 256x256.


Yes to most of this, but no to part of it. The yes part is everything you said about 256. For performance reasons, keep textures 256x256 or smaller as often as you possibly can. Only go bigger when you absolutely, positively have no other practical choice.

The no part is the PNG part. There's nothing wrong with using PNG as your source format, but understand there's nothing to be gained by it either. When the file gets uploaded, it is saved on the server in JPEG2000 format. You could source it from a PNG or a TGA or BMP and it will come out exactly the same.

I recommend using TGA over PNG because it's easier to prevent certain "transparency accidents" with TGA. But use whichever you're more comfortable with.



That's interesting. I had thought it was a female name too. My guess had been you were female, that you were into wolves, and that you were making a play on words by melding Cleopatra and Beowulf. Now that you've explained it's medieval male name, it somehow seems to sound a whole lot more masculine when I think of it in that context. Either way, it's a cool name.

and i recommend the opposite

why? personal preference, in my distorted little world i like the options of the much older PNG format (ie not saving color data on transparent pixels) ... which 99.999% of people wouldn't even know exist anyway

~and~

most other programs dont export TGA, 3d programs, vector drawers, and tons of other "non paint" software export 32 bit images ONLY as png

again it doesnt matter, cause as soon as you click upload its converted to jpeg2000 and sent over the ninernet to SL

Personal preference aside, Osgeld, where are you getting that PNG is older than TGA? TGA was invented in 1984, and hasn't changed since 1989. It's been a mega-giant staple of the graphics industry ever since, as the single most common format used for texturing and for all kinds of other things.

PNG, on the other hand, is a relative newcomer as graphics formats go. It first appeared in 1996, but didn't catch on as a standard until 2003, less than 5 years ago. It's only in the last few years that more than a handful programs have been able to utilize PNG at all.

I can't imagine why you think most 3D programs don't support TGA. TGA has been the industry standard format of choice for texturing for decades. While there very well may be some under-powered 3D software out there that doesn't support it, you'd be hard pressed to find a single full-featured professional grade tool that doesn't.

I also have no idea why you'd think vector programs don't support TGA. Illustrator has been able to import and export TGA practically since forever. As far as I know, the same is true for Freehand and Corel Draw. Sure, vector programs don't make active use of TGA. But they're not supposed to. TGA is not a vector format, and by the way, neither is PNG. Vector programs aren't going to use either one as a matter of course. The can import and export them if the need arises though.

Also, TGA has been a widely used standard in the video industry for decades as well. All full-featured video editing programs can make extensive use of TGA files.

In other words, TGA is hardly a "paint-program-only" format. It never has been, and never will be.


Anyway, as I said, there's nothing inherently wrong or bad about PNG. It's a great format. The only danger where SL is concerned is that since PNG supports both simple transparency and alpha transparency, sometimes the conversion to JPEG2000 gets botched, and you end up with a 32-bit texture in SL, where you meant to have a 24-bit one. It doesn't happen often, but it does happen. With TGA, there's no danger of that. A TGA-sourced JPEG2000 can only come out as 32-bit if you deliberately put the extra channel into the source file.

Also, as I so often talk about, understanding alpha mapping is crucial not just for transparency, but for countless other things in graphics. It's really one of the primary foundations of the whole thing. In that regard, PNG can be a little bit of a handicap for some people. If somebody only ever uses PNG's simple transparency, and never bothers to learn how to use alpha channels, they'll find themselves pretty severely crippled when it comes time to do anything that actually requires an alpha map. Since TGA forces you to learn that skill, it serves as a much better "teaching format", or "good habit forming format", than does PNG.

/me peeks back in. Sorry Cleo..didnt mean to gender bend you. I have no problems with being corrected. Some are teachers..some are..well...some should just do and not teach.

my two cents worth.

~Lana Tomba

it was adopted as an iso standard in 1995, the format just like the other portable graphics are some of the very first computer graphics ever dating back to the 1960's, over in the windows and photoshop world it just caught on, its been a standard in the *nix community well before 1995 becuase of its licensing, downside was you had to know how the file was written before it was ISO

well if i pay 5 grand for a pro package it better accept anything i toss at it, but were not in the realm of pros and 90% of all opensource or freeware software wont support TGA



maby you missed the word EXPORT (i know tga and png are not vector dont insult me) and again if i pay 5 grand for a pro package it better be wiping my butt with file options, but ya know what? regular joe smho tools dont export tga, png only in 32 bit, must be nice to live in your world with fully licenced photoshop maya 3ds max illistator corel and a gold sink to wash dishes in



only for professionals with an unlimited budget, again for the rest of us PNG is it



and ive never used it untill i got to SL where all of a sudden its the defacto texture image standard, ive always asscosaited it with gigantic floppy disk digital cameras, huge file size and poor quality for decades

I'll take your word for it on this since I'm afraid I'm not all that up on my ancient computer history of vacuum tube-filled rooms and punch cards. If in fact some distant ancestor of what we now know of as PNG existed 40 years ago, that's a really interesting tidbit of history, but it doesn't really change anything. If it's true, then it must also be true that all the ancestors of every other bitmap format (including those of TGA) must have existed back then too. None of that alters the fact that TGA as we know it today is signficantly older than PNG as we know it today. Nor does it change the fact that TGA has been a 3D and digital video industry standard for almost as long as both industries have existed, while PNG is fairly alien to both those fields, relatively speaking. I don't know why we're even arguing about this.


I can't speak intelligently to what percentage of all the underpowered open source freeware in existence might or might not support anything, including TGA. What I can say is that for anyone intentionally to make a software package that doesn't support the most commonly used formats for its particular genre seems to me like a pretty silly thing to do. I can't pretend to understand the mental workings of the people who write those programs though, obviously.

In any case, I'd be really surprised if it's actually true that such a high percentage of those programs don't support TGA. The most popular ones certainly do. GIMP always has, as far as I know. So has Blender. The only open source video editor I know of is Jahshaka, and it does support TGA, from what I've read of it. So what are all these programs that don't? Are they things most people would actually use?

Also, it's worth noting that the figure you twice quoted of $5000 for commercial software is quite an exaggeration. Photoshop has been the most popular image editor on the planet for the better part of two decades now, and while it's arguably expensive at several hundred dollars , it doesn't cost anything close to $5000. The same is true of Illustrator (although it's debatable whether or not it was always the most popular vector editor, even though it certainly is now). Heck, you can get the entire Adobe Master Collection suite, which includes every major program they make, for half that price, $2499.

Even most professional grade 3D programs like Max and Maya aren't in that sort of ballpark. At one time they were over 5K, but that hasn't been true for either one of them for quite a while now. (Well, Maya Unlimited costs more than that, but not many individuals have need of the extra features Unlimited adds to Maya.) So what are all these $5000+ programs that you think are the only ones that support TGA?

Of course, it's not that the above mentioned programs don't come with hefty price tags. To most people, a couple hundred to a couple thousand dollars is a good chunk of change. It's just not fair to keep shouting out the figure of $5000 like it's that or nothing. Let's keep it in the proper perspective, shall we?



No, I didn't miss the word "export". Maybe you missed what I actually said, which was that the three very popular vector editors that I mentioned can all import AND EXPORT TGA files.

And yes, I know full well that you know TGA and PNG are not vector formats. So there's no need to feel insulted; you haven't been. Please realize I'm not just speaking to you directly here. This is a public discussion happening in a public forum, so I'm speaking to everyone who's reading. Some people might not know what format is what, so for their benefit, it's important to mention it. Also, if I hadn't brought it up, you gotta just know that somebody would have chimed in with it anyway. So why not nip that in the bud? It's best not to leave the door open for confusion whenever it can be closed in advance, isn't it?




You certainly don't need an unlimited budget to use GIMP or Blender or PSP or Wings or any of the dozens of other inexpensive and/or free programs I could name that support TGA. Granted, professionals tend not to use those programs, but amateurs, hobbyists, and semi-professionals who want to do pro quality work absolutely do.

And for what it's worth, I've never had an unlimited budget, nor has anyone else I've ever known. In the beginning, I bought my software at student rates because I was in school at the time. Student discounts make this stuff incredibly cheap.

After that, I simply heeded the advice of a teacher who said, "Charge a deposit, and then use that money to buy whatever software you need to do the job." It's a great strategy. At a billing rate of $30-100 per hour, which is what most people charge to do this sort of work, one or two jobs can easily pay for any of the programs we've talked about.




I find that really, really, really surprising. (No offense.) As I said, TGA has been the most commonly used format for texturing and for many other graphics purposes for decades. Rip open any video game, and you'll find TGA textures all over the place. Take a course on 3D modeling or video editing at any school in the world, and they'll have you using TGA every day of the week.

It's precisely because it's such a common standard that TGA was even chosen for SL in the first place. Virtually all 3D artists are familiar with the common TGA work flow, so why not incorporate what people already know into SL? It makes perfect sense, and it was the right decision to make.

Where I think LL did goof was in not including other viable choices like PNG right from the start, but luckily that problem has been at least partially remedied by now. I'd still like to see it be able to import other formats like TIFF or PSD, which would save an awful lot of output work, but that's another subject.



I hope you don't still think that.

Just to explain, there's nothing "huge" about the file size. It's exactly what it's supposed to be for any uncompressed bitmap. In fact, the predictability of the file size is one of the things that helped make TGA such a widely adopted standard.

And it's certainly NOT "poor quality". By its very nature, an uncompressed bitmap is the highest possible quality image that can be displayed on a screen. Other formats (including PNG) can incorporate more bits per channel than TGA can, so in that sense, it's possible to kind of sort of define them as maybe "higher quality". But in the color space commonly used on personal computers, you're not going to see the benefit of those depths anyway.




The bottom line is this. If you like PNG, fine. There's nothing wrong with it. It's a wonderful format. Just don't make the mistake of claiming that TGA is somehow inferior or uncommon. It's certainly not.

Cleowolf, all other discussion aside, if you upload an image that is not standard SL recognised size, it will distort the image to suit, which is what a few people have already explained. All you have to do when you place it on a prim is stretch and resize the prim so that the image is back to it's original proportions. I've done it with much of my own artwork that is not the standard SL size and it works just fine

Me too. Depending on the detail of what I want to show as clearly as possible I simply resize my texture beofore uploading to that largest demension (1024, 512, etc) and let the smaller demenson fall where it may. Then size the prim I apply it on to undistort my texture. Most of my uploads are for clothes and building materials so I will always use 512 or 256 for those. But for snapshots I take I generally use the same resolution as my UI screen (1024 x 76 which is a 5 to 4 ratio........just maintain that ratio for your prim and there is no distortion what so ever. Same if I'm make a sign.....just keep the ratio of the width vs height in mind when designing and you never have distortion.

Another way to do it if you insist on not letting SL do it for you at the upload process is to go ahead and "distort" you texture before hand to force SL to let you have complete control on that process. There probably is a slight advance to that method because I could distort my 1024 x 768 texture to 1024 x 1024 first and then reduce the small dimension back to the 5 to 4 ratio.........sizing down is better than sizing up. But I sure can't see the difference when the texture is viewed on my computer (and I doubt anyone else could either). Either method works quite well.

The process is not really complicated at all.

That is absolutely ridiculous. There is no good reason for it to behave that way. Resize if it must, but to not constrain the proportions... what is this, 1982?

nope 2007, and the reason these limits are in place is to speed efficiency, and streamline the data, and generally go faster. it makes the input 'nice' for clean fast output...

... or to paraphrase Babbage, garbage in, garbage out

I get that... and after reading the guide linked in one of the first posts it makes a little more sense as I was viewing it from a web design perspective.

I see where it's ideal to reduce the texture memory footprint, but it still boggles my mind that there isn't a cleaner way to do it. I'm a photographer and I'm interested in putting my work in SL... these limits force me to do things to my work that I wouldn't normally do. Yes, I understand that I can allow the distortion and then set object dimensions to remove the distortion; but the fact remains that I'm still altering my work to some degree to even be able to get it to that point.

Maybe I'm just new to SL. Maybe I'm frustrated that there weren't clear instructions and I wasted a ton of Lindens in 2 bulk uploads before searching a forum for an answer that I feel should have been readily and easily accessible in the first place.

Bella, you're absolutely right that SL needs better instructions. Between the F1 help, the knowledge base, these forums, weekly video tips, orientation, third party books and websites, etc., all the information is out there. It's just that there's no consolidated instruction manual. I'm not sure there even can be.

I'd love to see a Maya-like help system in SL, jam packed with instructions and tutorials. But keeping it up to date could be a real problem. While something like Maya gets updated once or twice a year, giving its help-writers plenty of time to do rewrites, SL changes all the time. Certain basics, like this texture size stuff, remain constant, but there are so many subjects that don't.

Anyway, you shouldn't have to change the proportions of your photography. If a pre-existing photo doesn't fit easily into power-of-two proportions, simply mount it on a canvas that does. Then it's just a matter of using the repeat/offset settings on the prim surface to show only the part of the canvas you want to see.

For example, let's say you're photo is 2400x3000 to start, typical resolution for an 8x10. Downsize it so the largest dimension (in this case the height) is a power of two, say, 1024, and the smaller dimension will end up being 819. Put your 1024x819 photo on a 1024x1024 canvas. Since you know the width to height ratio is 8:10, and since you know the height of the photo is filling the whole canvas, you know that when you apply the texture to a prim surface, you'll want to set the vertical repeat to 1, and the horizontal repeat to 0.8. By showing less that one full repeat for the width in this way, you "trim off" the excess blank canvas from view. The result is your 8x10 photo looks perfect in SL.

You waste a small amount of texture memory each time you do this, but generally it's worth it. If you do it the other way (reproportion the photo to fit, and then stretch it back out by stretching the prim), you end up losing a considerable amount of quality from the image. By using this mounting technique, you preserve as near to full quality as is possible.


On a general note, it's important for people to realize when they get into texturing for the first time that texturing is not photography, texturing is not web design, texturing is texturing. It has its own rules. Were you putting your imagery into a video game or any other realtime 3D environment, these same rules would apply. Just as every photographer has to learn that images meant for the screen have different rules than images meant for print (different color spaces, different dpi's, etc.), anyone who wants to do 3D has to learn that images meant to be textures have their own rules as well.

It's no different than anything else, really. Non-digital artists have to deal with "rules" also. If you're going to do an oil painting, you have to do different things to prepare than if it's going to be an acrylic painting. If you're doing a charcoal drawing, you'll do different things that if it's a pencil drawing. You don't see to many artists out there saying, "How come nobody told me I couldn't use watercolor to paint my motorcycle tank?" It's just assumed that people would know better than that.

Digital media are no different. You must expect to have to learn new rules and techniques for each new medium you approach.

Hey Chosen, I am sorry if this has been answered before but where or what are the rules?
Can you point out where to go read up on these "rules" to those of who us who are new or just forgotten or don't know.
Thank you, FD

By way I have learned the hard way if I work on image that for example is 2048 x 2048 and I reduce the image in my graphic program and upload it to SL some of image seems to lose
it's detail and often turns blurry. I noticed this also when I was trying to do complex window scenes on 512 x 512 size that I had divided the images into ten 50 x 50 scenes in between a small border areas in one texture.
I am not sure though how to fix this.
Any suggestions?

I don't know what proggy you use to make your textures..but in The GIMP theres a little feature under "Filters"->"Enhance"->"Sharpen".

I've figure out an average of 55 on that slider in sharpening..compensates for any blurring that sl might distort your image into.Also if the original resource file is not "clean"pixelated or blotchy) due to a small file size or jpeg format..adding a gaussian blur after scaling the image (for the most part cleans up that pixelation). make sure the blur is set to 1.0 -not the program preset of 5.0.

Then i go back in after blurring the pixelation and add sharpness in incraments until I get my desired result.Usually works...(even on sad little jpegs)

~Lana Tomba

P.S. oh..sometimes..a slight contrast adjustment of about 0.34 add just enough enhancement as well to make up for SLs funky glitches.Don't know if any of these terms are universal I'm sure theyre not.

I am using a older verison of Adobe CS version 8.0 and there is Filters; enhance/sharpen
but it seems sometimes to have more destructive effect yet sometimes it works too.
What do you mean Lana by a slight constrast adjust of about 0.34?
I don't entirely understand.
Thank you for responding Lana.
FD Spark.