Adding to Cart…

Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
I guess people worry too much about disk space? No idea why use JPG for texture maps otherwise.
------
Here is my take on the free beholder for Poser ( http://threednd.com/monsters.html ) - the eyewhites could've looked better, but the sclera map included seems to be the wrong UV, and I wasn't in the mood to paint a new one. The "lens" (which is anatomically cornea, but called "lens" in the surface list) of the primary eye doesn't play well with sub-d or refraction due to the way it's modeled, so it's a simple reflective overlay.
But I forgive any faults because it's a free beholder model!!
Two path-traced area lights, a GI envlight with one of my Fantasy maps, beholder surfaces use SSS, the Dungeon Corner surfaces do not. Rendered to EXR, tonemapped in Picturenaut.
And pngs aren't much, if any bigger and very often don't have the nasty artifacts that jpgs can...that's if you must go the compressed image route.
Besides, when it comes to usage, the renderer uncompresses them anyway, so there isn't really much, if any savings in texture memory (at least Iray does, unless you have it's compression down at 512 or something...it's going to use 3 bytes/pixel whether it's a tif or jpg...and tdlmake is doing some sort of compression manipulation during the conversion process). And yeah, tifs can be compressed...but to me it's a 'why would you' question.
I did a series of render tests back in 4.5 or so, uncompressed tifs actually rendered a few seconds faster (not counting 'optimizing image' time..it was always in the 5-10 second range, which was just outside the render variance) and were 'better' looking, even at smaller resolutions than jpgs. A 2k tif looked as good as or better than a 4k jpg. Lightly compressed pngs and bmps were tied for second place and jpgs were the worst looking/slowest... I couldn't figure out why, after being optimized it would have an impact on render time...but the finished 'look', yeah, that was easy to understand.
If I had to chose, I'd say png over jpg...but tifs would be even better.
PNG are the best for the "non-technical" users, I'd say. TIFF is good, but isn't EXR even better?
I keep meaning to check if tdlmake understands that it does not need to process EXRs (because 3DL handles them internally).
I haven't looked either...but my guess is that even if 3DL will handle them with no problems, Studio will pass them to tdlmake to do something with anyway. And yes, EXRs would be even better...but most of my working files are not exrs.
Textures fidelity depends on a lot of things, not just compression format. You can have a lossless JPG if you want (quality at 100%). Or use the newer versions JPG2000, JPG-XR that's generally better at lower quality settings. My guess is that most vendor just used whatever default setting their image editor uses (typically 90 or 95). JPG at 100% Quality is still smaller than PNG, with virtually no compression artifacts.
Of course, it's still limited to 8 bits per color channel. TIFF and EXR are better because they can use higher bits per channel. You generally would need that kind of fidelity for other than color maps (normal, bump, displacement etc).
I'm not sure that all JPG encoders will produce a true lossless JPEG at 100% or support any of the more recent updates to the standard.
As far as I am aware even at 100% quality a jpg is not lossless - I remember extended "discussions" of the issue on the Photoshop forums.
Yes, my mistake there. I should've wrote, almost lossless.
I did wrote 'virtually' no compression artifacts. There's plenty of compression dials you can tweak to get very close to lossless (quality, DCT method, smoothing factor, subsampling). They help a bit, particularly when coupled with higher resolutions (more pixels).
Remember that JPEG compression is a perceptual, lossy compression scheme. They are not really suitable for anything other than color compression. Plus it doesn't solve the format lack of bits in the color channel.
JPEG2000 offers lossless compression though, plus higher bit depths. But of course, everyone just stuck to the old JPEG standard when saving.
Test renders of final wood materials. Texture is from here - http://webtreats.mysitemyway.com/category/textures/
That wood looks cool, Wowie; does it use any SSS at all?
As for JPG, I haven't yet saved any textures to it, but purely for viewing purposes, I find that disabling chroma subsampling makes for the best-looking image - most crisp. I use RIOT: http://luci.criosweb.ro/riot/
Here's a screenshot: http://dist.alternativeto.net/s/3761b194-7134-4b01-b532-02c99415c193_3_full.jpg?format=jpg&width=1900 - easy to use, I think.
My preferred settings for web versions of renders/scans (the minimum size which is the closest to the original, to my eye): no subsampling, 90%, progressive (helps reduce the size further)
For instance, if you take my userpic - http://mustakettu85.deviantart.com/art/Against-all-odds-277837871 - even the full version looks like crap with any chroma subsampling (the red carnation turns into a blur). Just try resaving it and see =)
Nope. No SSS. Just basic diffuse/clay roughness.
Yeah, I've tried RIOT. I don't use it though. I generally just use Xnview for viewing and sometimes converting. I generally just use 95% Quality with the same subsampling as you (1x1), with 0 (zero) smoothing and floating point DCT in Xnview. For GIMP, it's 100% quality and pretty much the same for everything else. Still wish they hurry up with the GEGL and high bit per channel support.
Btw, this is pretty interesting. Should help regardless of your renderer.
http://www.innobright.com/
It's a denoiser. Of course, it needs AOVs, which unfortunately is rather tricky with DS.
Yeah, I'm waiting for HDR in GIMP, too. Krita is fun to paint silly light maps in, meanwhile. =)
The denoiser looks promising, but as one of the guys said in this thread ( http://blenderartists.org/forum/showthread.php?384931-New-rendering-specialised-noise-reduction-Altus - specifies the necessary AOVs, too) , the real test will be a scene with a lot of hi-frequency detail. The hair example on the website looks a bit "painted" to me.
Krita is fun.
Some stone/ceramic-ish presets.
So cool =)
Maybe it could be worth it to re-position lights so that Fresnel would be more prominent on the specular, especially in the first render where the surface is the roughest.
Speaking of Krita:
https://krita.org/item/krita-2-9-animation-edition-beta-released/
Hmm, interesting
http://renderstory.com/fast-renders/
From memory the Nvidia plug in for Photoshop allowed saving DXT3 and DXT5...and was free. Would probably work with Gimp also.
Interesting thread...when my head isn't so fuzzy I'll have to read it through from the beginning.
Sorry, quote is showing up wrong and I can't seem to change it...
The quotes do take a bit of getting used to get them working just right...
But yeah, I think it was the Nvidia tools. And there is a GIMP or was a way of running it in GIMP. But the funny thing is, I always have an easier time getting dds to work in my Linux GIMP than my son does in his Windows version.
I've always used it in PS as I had a student version from a long time ago. I keep meaning to give GIMP a go as it's on the computers at school.
There's a lot GIMP can do...but there's a few things it can't...yet. And depending on which version of PS you have you may not notice that many missing features.
Welcome to the lab Pendraia. The forum sometimes dose that goof up, Images don't upload, Past text without formatting still puts in a link (It just did that three times with your name, as I was pasting it in), Comments don't quote quite right, etc.
wowie, adding to. JPG has only a 256 color depth if memory dose not serve, lol. (GIF not JPG, sorry) Isn't that essentially 8bit color in a 24bit sRGB world, artifacts would be the least of my concerns with gradual shading in allot of stuff, lol. There is a considerable in depth review of it somewhere (Wikipedia I think, possibly, somewhere), I had read the article at least fifteen years ago or there about.
http://en.wikipedia.org/wiki/GIF
http://en.wikipedia.org/wiki/JPEG
Here is the articles,
http://en.wikipedia.org/wiki/Image_compression
http://en.wikipedia.org/wiki/Lossy_compression
http://en.wikipedia.org/wiki/Lossless_compression
http://en.wikipedia.org/wiki/Compression_artifact
And JPEG2000 (JP2) costs money to get things to be able to understand it (Or at least that's what I understand from the warning in IrfanView, when I try to save one on occasion by accident). If the standard was not under such strict user rights restrictions, It probably would be more popular then the older 8bit JPG format.
I've started doing as much as possible with PNG (Portable Network Graphics), however sometimes limits on image MB often forces me to go to a format with Lossy Compression (jpg, gif, cringe), especially on the forum.
http://en.wikipedia.org/wiki/Portable_Network_Graphics
(EDIT) Note on IrfanView. I have stopped looking at newer versions, as everything after 3.98 has resorted to odd image resizing when you resize the window, and the zoom thing also changed drastically (I hate the new method, others Love it). So if the restriction on JPEG2000 has been lifted, I am unaware of it, and still can not use it.
Hi Pendraia, and welcome =)
Yeah the official Nvidia Photoshop plugin did have everything... but Photoshop itself isn't free, and was not that easy to buy in Russia back then (2007), to boot (there were a couple resellers, I think, but no student discounts or whatever).
I was on Vista back then, and GIMP plain out refused to run. CTD on launch. What I had was some early Paint.NET plugin (the one that wouldn't crash LOL). These days the PdN plugin looks to be more robust.
Zarcon, JPG is 8 bit per channel (red/green/blue), that is, "24bit" in 'computer screen colour' speak because 3x8=24. Those 8 bits can encode 256 steps - which means 'true colour' because 256 to the power of 3 gives approx. 16 million combinations (aka colours).
But it's not the same thing as HDR which has 16 or 32 bits per channel.
Sorry, Perhaps I mixed up GIF with JPG, sorry. As for the rest, that's my list up there. If it is not in that list, I can't use it unfortunately, lol. Perhaps TDL make can, good for it, it dose me no good, lol. Not in that list, I can not create, edit, or view it, lol. and as shown, there are limits on the use of some of them.
Pendraia. Oh, and some times you need to go in a click the Spell check a dozen times before it actually dose it, lol..
IrfanView can view a lot more than it can save in.
Tdlmake is strictly 3Delight-only utility. You don't author files in it (outside simplest skies), it's just a converter.
To make meaningful 16bit/24bit maps, you need programs that can generate HDR data. The simplest example is rendering a sky panorama in Bryce and saving as HDR for use in other programs.
PS For those familiar with HTML syntax, there is the "source" button in the post editor. Very handy.
I'm not sure...they update the image from time to time and I would have to check on the school computers. My new one doesn't have a lot of the programs on it. I'll have to see about getting it updated. They even have Blender on the image funnily enough.
I fully understand...as I said earlier I was lucky enough to have a student edition otherwise it wouldn't have been affordable. Elements is cheaper and the plug-ins work with it also but it is missing a couple of features in the version of elements that I bought.
I haven't tried PdN but I'm glad that it works better now. Must admit I like that you can create texture sets for Skyrim which makes texturing much easier.
Thanks for the tips Zarcon...sorry if I took the thread off-topic.
This is the 3delight Lab, this is still part of that, and thus I think still very much on topic. Render methods, shaders, even texture formats.
I added a few links to the former Post, and I think that one screen-cap is all that is needed to explain JPEG2000 (JP2) formats, lol. I was just throwing in my two pence (something smaller then a penny) on the topic.
(EDIT) So, My thoughts is not one of one or the other must be used all the time for everything. It's knowing where each method is best, and using the compression method that is best for the use that you are using it for. There are times you do want the best smooth transitions for things like a cloudless sky, and others when it is just a waist of disk space like a Thread level cloth bump map (that is going from max white to min black in less then eight pixels across the map).
I'm too "oldschool" for Skyrim, can't seem to get into it at all =) But I did some personal recolours for The Witcher textures recently, and they are doing fine, mipmaps and all.
I found a test I forgot to post here... this is a mix of my shaders (for the metallic cones), lights (GI etc) and DS Default shader with various degrees of ambient on - faking SSS on the hexagons, and making the orange spheres a bit emissive (you can't really tell, though).
The spheres look like streaks because they're motion-blurred. The whole mess is a test of Tofusan's surface replicator over a random shape whipped up in Scultpris.
Important: The key to get motion blur on instances is to move the original instanced mesh, not the instance group.
Well, in a production environment (in a VFX studio, for instance), AFAIK everything will be authored to the max quality possible.
Another application when hi bit range is indispensable, is large-scale displacement mapping. Casual had very descriptive screenshots here: http://www.daz3d.com/forums/discussion/3269/
For instance, when you use a program like Allegorithmic Bitmap2Material to convert normal maps to displacement maps, the best way is to save the displacement to EXR.
Also, 16 and 32 bit formats...like exr are linear by default...there is no need to mess around trying to figure it out. And having one set of maps that is usable in ANY renderer is very desirable. And I wouldn't be surprsed when in a few versions 3DL will require a linear workflow.
This, too.
I hope we'll get a switch to disable tdlmake in DS, then =) I'd rather find something that can save mipmapped EXRs and feed them to 3DL raw.