tga v png

I agree, Johan. It took me a while to figure out why anyone was having halo problems. The way I work, I always have a layer of fabric somewhere that fills the entire canvas. It has never made sense to me to cut out fabric pieces, since the alpha channel is essentially a pattern -- it will do all the "cutting" for you. If you leave the fabric layer alone and let it fill the canvas, it's impossible to get a halo. Chosen and Robin referred to this approach, when I described it in a post last Fall, as the "subtractive method" and suggested that it was hard for a newcomer to understand.

It seems to me that quite the opposite is true, especially if you have ever done any sewing in RL and know how to work with patterns. Thinking of the alpha channel as a pattern helps you avoid the trap of confusing it with a layer, and it certainly helps you avoid all the fuss and bother of removing a halo that you didn't need to create in the first place. I just tell myself, "Layers are for painting; the alpha channel is for cutting." (Not strictly true, but it's a helpful mantra. )

Looks like I opened up a can of works here. Let me try to straighten this out.

Trust me, guys. I've been teaching this stuff for a long time. People like yourselves, who have a relatively easy time understanding the concept of alpha channels right away, are rare. Most people are not well able to think in those terms without a good amount of experience under their belts first.

Johan, while you're certainly right that it's never a good idea to steer people in the wrong direction first, and then tell them to pull a 180 later, I can promise you that's not what's happening with my tutorials. The way they're designed is to walk people through processes that naturally put them in position to grasp the basic concepts the most easily. I'll try to explain.

The way most people naturally approach artwork is to think additively. For example, when most people draw a picture, they'll they'll start with a white piece of paper, and add blackness with the pencil to create shading. Sooner or later, those who have a very good understanding of light and shadow will make the leap to realize that it's often a much faster approach to start out by blackening the whole paper with charcoal, and then drawing subtractively with a good eraser.

In traditional art classes, subtraction is not taught until students are already very good at addition. If art teachers were to try to start out absolute beginners with subtractive methodology, 90% of the class wouldn't get it, and would likely never learn to draw. Even among those with years of addition under their belts, I've seen many a talented artist struggle for weeks when confronted with subtraction before they're able to make it work for them. Some are able to transition back and forth with complete ease (I certainly was, and it sounds like you guys might as well), but not most. For most people, it's not just an alteration of technique, but an evolution of thought process, which takes time to develop.

For every person who might thrive by being taught subtraction first, whether we're talking about traditional art or digital art, there are at least nine more who wouldn't yet be equipped mentally to deal with it. This stuff has to be taught in steps. I know it can be frustrating for those rare people, who can just get it on their own, to have to say, "Why didn't you tell me that in the first place?" But anything presented to more than one person at a time must be put in whatever order that the greatest percentage of people will be able to understand.


For a comparison, it's a bit like learning a musical instrument. When teaching the guitar, you don't start the student out with bar chords. 99% of them just won't have the muscle strength to be able to play them yet. You start out with open chords. And even then, most teachers won't begin with the full six-string versions. Usually just three-string versions are taught at first. Only once the student has mastered those does the teacher say, "OK, now let's expand on that. Put the rest of your fingers here, here, and here." And not until weeks or months later are bar chords even mentioned.

Every once in a while you get someone like Steve Vai, who from the very first day he picks the thing up finds there's almost nothing he's not able to do with it. (I once read an interview with Joe Satriani, in which he said, "Teaching Steve was like teaching a robot. Anything I'd show him, he could just do. His fingers were like machines." But again, those people are extremely unusual. It just wouldn't make sense for teachers to try to teach every student in the way those rare individuals are capable of learning.


In the end, everyone is capable of learning subtraction, just like everyone is capable of playing a bar chord. Both are very simple tasks, when viewed in retrospect. But for most people, it's anything but easy at first. Just as finger strength and physical coordination need to be developed before the playing bar chords is possible, a certain mental strength and visual/conceptual coordination has to be developed before subtraction is possible. I hope that makes sense.

Yeah, I can see that, having played the guitar for quite a few decades myself and having fiddled as an amateur artist at least as long. Perhaps the reason that I had no trouble with alpha channels from the start is that my mind was not on art but on dressmaking. From that perspective, it makes sense to begin with a full, uncut piece of fabric and then work with a pattern to shape it into the pieces you want. The alpha channel has always been my pattern. If I had been unfortunate enough to start by thinking of clothing design in Photoshop instead as a drawing exercise, I might have taken ages to have an "aha!" moment and understand what I was doing.

What it comes down to is a framing problem. The way you conceptualize a challenge from the beginning determines the kinds of tools you choose to meet it, and the methods that seem most logical. A successful learner (or a good teacher) is often one who can stand way back and understand the frame before diving in to address the challenge.

This is not technically accurate. There are formats where a set of pixel values denote "transparent" rather than a color. This can be done in 24-bit mode, or smaller bit width when using pallettes. You can have also a pallettte with number of different semi-transparent colors.

The term "alpha channel" implies an extra 8 bits of transparency per pixel, allowing the transparency mask to be as complex as any color plane for the image, allowing images that would require an infeasibly large palette.

You may be right that for PNG the difference in how transparency is encoded is not significant; I wouldn't know. In any case, images are stored internally in the same format, so once the conversion is applied, they all behave the same, and to that extent your remark is correct.

I don't know much about PNG format; all I know is
(a) it does support transparency in SL, and
(b) when I try to use it for skins, it doesn't work, for reasons unknown, even though the images preview identically to the TGA counterparts in various applications.

I was in the habit of using PNG for alpha textures for objects, since Windows thumbnails works for them. Someone pointed out a simple package that adds tga thumbnails to Windows, which was great.

Once you've mastered barre chords, what do you spend the rest of your life learning? *Not* to use them. Barre chords form an important stage in the learning process, but they're used sparingly by most good rhythm guitarists.

They serve a number of purposes, but the main reason they're used is that they're relatively simple and systematic; easy to grasp a handful of shapes to allow you to play hundreds of different chords, while familiarizing yourself with different positions on the neck.

Barre chords are the additive part of learning to play guitar. Many jazz chord voicings are simply these barre chords with several of the notes removed, and the good jazz player is aware of the notes not being played. This applies equally to rock; rock power chord voicings leave out key notes and reinforce others. Same concept, applied differently. In either case, few musicians are gifted enough to learn the notes to leave out without first learning to play them.

That's strange. There's really no reason it shouldn't work. What exactly happens when you try?


I had to read your post several times to figure out how it was relevant to the topic at hand. But if I'm interpreting you right, I think the above sentence is the crux of it. It's a pretty good way of putting it, I think. Tell me if I'm wrong, but I think what you're saying is that in order to know how to subtract most effectively, you must first develop a solid understanding of the whole, and starting with addition is an effective way to learn that.

That's certainly one good way to look at it. A more fundamental interpretation would be that you need to develop an understanding of the basics -- of drawing, sculpting, writing, music, whatever -- before you can improvise. In theory, it shouldn't make any more sense to gain that understanding by an additive approach than by a subtractive one. What's important is that you reflect on the principles behind the approach, rather than simply the rules.

BTW, after 45 years, I still use barre chords when they suit my needs.

Yup, and just riffing on your own simile. You mentioned barre chords as the thing to avoid at first as too advanced for beginners, but it turns out it's a better metaphor for the additive methods in two ways. Not only is it *not* the goal but merely the path for relative novices, and as you say, it helps you develop the understanding of the whole before removing that which should not be there.

There's another nice little metaphor for subtractive work. I don't remember who said it, but sculpting is the art of finding the statue in the stone and removing the bits of stone that aren't part of it. But those who would be sculptors must first learn to paint.

Whether this has any bearing on learning PSP, I leave to you texture wizards.

Of course. I said "sparingly", not "never". I use them more than I'd like, but when I watch the guitarists I admire most, they use them far less than I do and pretty much regardless of genre. Of course there are genres like grunge where they predominate, possibly as a rebellion against sophistication. Which is cool too.

IIRC, the alpha simply doesn't work, or works wrong. So, it looks fine for legs and body, but the head is all messed up, especially eyes and eyebrows. It's possible I did something wrong each time I did this, but the TGAs saved from the same source worked fine. I'm no skin maker; this was mostly just fooling around.

I use.....

PNG 32BIT: Only with simple transparency such as the wreath
PNG 24BIT: For non transparent textures
JPG: For most simple non transparent graphics.
TGA 32BIT: For complex textures such as window glass that have frames or doors.

ive seen many file extensions come and go, its not funny anymore.

I would like to make a comment. I hope no one takes it wrong. It is not meant to be insulting to anyone. Only meant as a simple opinion from someone starting out and using/needing tutorials.

For some of us, we simply want an easy hands-on, step by slow step, beginner How-to tutorial. We don't care about the technical information or the why it is this way or not that way. We just want to follow the steps and make something.

Some of us know nothing of what is taught in formal education settings for things like art classes, computer programming, in depth photo shop usage, etc. nor do we want to. Some of us just want to put our creativeness into play. Some of us learn by doing. We go cross eyed at all the technical stuff or extraneous stuff that has no bearing on the how to...we are not prepared to assimilate and use yet. We pound our heads on the keyboard trying to sort out all the stuff from the steps of making something.

Once we practice the How-To and become more confident we might venture deeper. We might meekly post questions on the forums asking a question for something we don't understand...if we become really confident we might post an opinion or answer to someone else's question. I know there was something posted earlier on the Forums that I have a question about, but I won't ask it cause people with think me stupid. And, I am far from stupid just I have limited information on the subject and am learning. I don't want to look like I am an idiot in front of you people.

I guess what I am suggesting here is 1) keep tutorials simple and project completion oriented; 2) keep the extraneous information to a minimum or a separate section; 3) create, beginner,novice, advanced tutorials; 4) keep tutorial subject matter separate (example, don't blend gimp in the same tutorial as photoshop or paint shop...it creates unnecessary confusion).

Yeah, too bad the forum won't let you search for anything under 4 letters, so searching by file type is usually pointless.

All excellent suggestions. However, I'd encourage anyone who has the ability and inclination to write a useful tutorial to go ahead and write it, even if it doesn't meet all these ideals.

I'd like to see more tutorials planted in the SL Wiki, because they can link to background or additional material, and everyone can contribute (yes, I realize there's a downside to this too!)

That's a decent observation, Morgaine. Sometimes, you do need to just get on with it. At the same time, I think a lot of people create problems for themselves unnecessarily because they look for quick cookbook answers instead of spending a little time to figure out how things work. There has to be a happy medium somewhere.

Since this thread is about alpha problems, I'm thinking in particular about how many people dive right into making things with transparency without taking a half hour to read about what an alpha channel is -- maybe going to some excellent site like Robin Wood's that walks through basic theory painlessly. The theory is IMPORTANT, and it's not all that hard to understand. I would guess that 90% of the questions we get in this forum about transparency would vanish if people read about it a bit before searching for some how-to-make-a dress-in-ten-minutes tutorial instead.

Again, like you, I'm not trying to be insulting to anyone. I get impatient and want to build neat things in a hurry too. I just hate to have anyone brush off theory as if it didn't make a difference.

That's a decent observation, Morgaine. Sometimes, you do need to just get on with it. At the same time, I think a lot of people create problems for themselves unnecessarily because they look for quick cookbook answers instead of spending a little time to figure out how things work. There has to be a happy medium somewhere.

Since this thread is about alpha problems, I'm thinking in particular about how many people dive right into making things with transparency without taking a half hour to read about what an alpha channel is -- maybe going to some excellent site like Robin Wood's that walks through basic theory painlessly. The theory is IMPORTANT, and it's not all that hard to understand. I would guess that 90% of the questions we get in this forum about transparency would vanish if people read about it a bit before searching for some how-to-make-a dress-in-ten-minutes tutorial instead.

Again, like you, I'm not trying to be insulting to anyone. I get impatient and want to build neat things in a hurry too. I just hate to have anyone brush off theory as if it didn't make a difference.

Hhhhhmmm...maybe I mispoke or gave the wrong impression. It is not that theory is unimportant, but for me a more hands on approach is how I learn best then I can go back and understand the theory and what it is trying to tell me. Most of the time, at least for me, I do read the theory but often times there is so much theory that it gets confusing and becomes a total turn off. It ends up sounding like eye glazing tech speak. Totally intimidating.

I have read the alpha and transparency theories several times over the past couple of years and still find them confusing, thus the earlier post on suggestions for future tutorials for folks that might learn as I do. This confusion put me off the subject for a long time till I recently found a tutorial that was interesting and well-written what walked me through the steps. I may not have a very good grasp on the theory, but was able to make my first successful alpha texture using .png. With that accomplishment, I gained some confidence and I am trying to learn more of the theory. And wow, I may even go back now and try making an alpha using .tga.

I guess I get impatient because alternative learners learn differently and are often brushed aside because we do have to take a different avenue to learning theory. I think all of those posting to the forums do a great job, sometimes a little impatient with some of us, but all and all I think people want to help others learn. And, I for one appreciate it.

You can say that again, Rolig!

I've found I am using both since upgrading PS.

If I need a trans I use the png with the alphas, if I need a solid texture I use tga. I find the tga at the lowest bit removes the bleed through on prims. (you know where they turn inside out, or go in front) tga with a higher results in said bleed throughs (sorry can't remember what its called exactly)

works for me.

Just to add to Morgaine's comment, (I know it has nothing to do with tga V png).

Sometimes giving a step by step instruction on how to do something goes a long way beyond the intended reply. eg A quick answer by Chosen on how to make a glass texture, and another link by someone (you know who you are) on how to make silks, sudenly means the lightbulb turns on.

Ahh thats what filters are all about !!

6 hours laterr when I have tried every filter in PS to see what it does, I feel my knowledge has increased considdrably.

Please don't underestimat the effect thes simple (for you) walkthroughs can have on us beginers. If it opens up some understanding of how things work it is effort well spent.

TGA with the "lowest bit" (by which I think you mean a TGA image saved as a 24-bit file instead of a 32-bit file) has no alpha channel and therefore no transparency, so it's no wonder that you have no alpha sorting issues with it. If you are using PNG for your transparent images, chances are that you are using simple transparency, which also has no alpha channel. Using PNG, it is also possible to create a 32-bit file that does have an alpha channel.

Regardless of which file format you use, the sorting issue should still be there, I believe. Chosen or someone with greater technical wisdom can correct me if I'm wrong. The sorting problem is not specifically an alpha issue, but a transparency one. When your viewer goes to draw a scene, the graphics card normally has no problem figuring out which objects to draw "in front" and which ones to hide "behind." When it has to deal with two transparent objects, however, it can't figure out which one is nearer, so it flips a coin and tries to draw both -- hence the flicker. The problem isn't the presence of an alpha channel per se, although we usually associate it with alpha channels because that's how we traditionally create transparency. It's really the transparency that is the culprit.

Overall, I'd say that if you haven't had sorting issues with your PNG-created transparencies, you have been lucky.