CompuServe Thread

I beleive that "100% quality" in JPG doesn't really mean to convey "no loss", but rather, that it is the "least-lossy" of the possible amounts of compression — so while you may _not_ see any compression/loss artifacts in some images at 100%, you _may_ in others. Dave

This is only from my experience and not based on any text book theories or formulae. The problem with using JPG, even at full quality, vs. targa, even RLE compressed targas is when you have to go through several codecs to reach your end product. When you store the jpg and then open it up again you go through one codec process. If you then use that jpg in a work and render it again, saving to a new jpg file you have two codecs. The affect of this multiple codec can be severe, much worse than true lossless compression such as RLE. I have found that the name of the game for lossless is to only use one codec in the project chain which should be the final phase. (codec: compression-decompression)

Depends on what you are doing with your final output. 100% is really only a best quality compression. JPEG does support a lossless saving, but I don't think I've ever seen that implemented. 100% JPEG images look darned good in most circumstances. There will be image degredation and artefacting. If you are just creating images for display on a monitor or some other straight use of the JPEG, then you are probably fine using it. If you expect to be doing further image manipulations, you may not be OK. Repeated JPEG saves will degrade the image to the point where you notice. Certain image enhancement operations can highlight the JPEG 8×8 pixel grid and other artefacting. But at 100% you might be quite OK. The best way, is to run a couple of tests. JPEG is much better at photo type images than at flat color and line cartoon like images. If you change your rendering type/look, then run your tests again. Note also that one of the first things you may notice in a JPEG image is not edge artefacting, but a change in color. – gruhn

My only comment is that you can't have Alpha information with JPG, so I would always use Targa. Alpha may not be a factor, but it's something to remember (I use it a LOT).

Garry, >> a 100% quality JPG should be as good (in quality) as a Targa – << not really. In 3ds even 100% quality is around 10:1 compression at which point there are plenty of artifacts and image fidelity loss. To be considered visually lossless, it seems the concensus is 3:1 or 4:1 compression is the right ratio.

I only use Jpg for two things, compressing a picture that will be used as a texture in a not so prominant place, and uploading to this forum. Maybe my eyes are just really waked, but I can tell immediately if something's been JPG'd, Just look at an area in between two distinct colors, you'll see strange out of place dots. Always make sure you have at least a tga backup of anything you Jpg. Jpging something twice is a big no no for me. – Neil

It depends on what the image you are working with is. I did an animation that had a catheter wire going through blood vessels. Because I was lacking hard disk space I rendered to JPEG. The wire was thin and black against a solid red. Even at the highest quality setting the wire came out looking fuzzy, as if I applied a 2 pixel gaussian blur to it. Unfortunately because of a tight deadline, I couldn't go back to Targas. Now I use mostly Targas. Compression is funny in that it is very particular to the image you are dealing with. I would do a test render with each and let your eyes be the judge.