hello, today I did little test and I am very disappointed to be honest. I always thought that when I save renders in 16bit tiff, I have greater color range to play with, but I think I live in lie…here you can see comparison between 16bit vs 8bit tiff + 16bit exr(REAL 16bit)…I found results to be exactly same between tiff 16bit and 8bit, both look like 8bit to me, well at least my eyes can not differentiate them no matter how much I change exposure…so my confusion starts here, why would anyone ever use 16bit from blender? in any format png or tiff, my tests show they look exactly alike under exposure changes…on EXR behaves how I expect to. and toll if high, tiff 16bit is twice size of tiff 8bit, question is where is those extra data?..my head ready to burst…
please someone who knows whats what, explain to me why would I ever use 16bit tiff/png??? or why is it even there? what is does? thank u
(am not talking about baking, only renders saved to 16bit tiff/png)
Additionally to the link thorn sent, this is the difference between an 8-bit and 16-bit image from Blender (heavily increased contrast so the difference is visible):
what I find is that, when I did that on 8bit tiff, yes there is bending okey, but as soon as I switch mode to 16bit from 8bit in photoshop, bending dessapears…altho file is same 8bit.
so if I use 8bit file and work in 16bit space(Mode) file behaves like 16bit…well in this case anyways…
my main dissapointment is how blender saves 16bit tiff/png. I always though that it saves more than what I can see, so I can bring back highlights for example but NOO. from now on I know that if something is cliping, 16bit wont save it…it saves more colors in between whats cliping on 8bit file, bellow 0 and above 255…that pure sadness…
basically 16bit only makes seance on height or normal maps, if u want to edit your renders, u should use EXR 16/32bit…or go with 8Bit tiff, no point for 16Bit at this point
If that works for you, that’s great - go for it. But I wouldn’t make it general advice. 32bit EXR is probably one of the best formats to render to, but it takes space and you also have to deal with color management, because it doesn’t save color management transforms in it which is often great as well, because you may want exactly that, but sometimes you do want the color management saved so TIFF format is then great for that. As for 16 vs 8 bits… Hello and welcome. Glad to see you starting with this. Maybe just slow down with your recommendations for others a little bit. 8 bits will produce all sorts of problems when editing really quickly so your advice is not a particularly good one.
I came here with questions, am just saying what am thinking. I did not gave anyone advice. we are just exchanging thoughts.
I am average blender user with basic color understanding and here I fall this delusion that 16bit saves more highlights/shadows than I can see in viewport. I bet you that many other blend users think that way (or am just especially stupid, am not ruling that out) anyhow. thanks everyone, I did learn something today.
but Advice will be now:
anyone who renders in blender(am not talking about Baking!!), know that saving 16bit tiff/png will clip same highlights/shadows as regular 8bit would. if u want broader range of colors use EXR format and do not worry that colors will be clipped. do not waste space on tiff/png 16bit files.
You have to understand a few things about EXR. It is for saving scene linear color space. So for example if you are using AgX, you will lose that with EXR and will have to recreate whatever color transform in the software you use for editing. Photoshop 2025(I believe I see you using it) now has some OCIO support - that’s great(haven’t tested it yet personally) and there is a plugin for EXR which makes other things possible as well(like Cryptomatte), but you need extra understanding and setup to use EXR with Photoshop and other software as well. Recovering or working with highlights is not only about clipping and you may want to shift colors/tones in your image in all sorts of ways while editing. 8 bit color depth may screw things up big time when editing for printing or in many different circumstances. Don’t just make a rule for yourself like this. It’s a better general rule to always use higher bit depth for any editing. It does depend on the circumstances though.
What I am trying to say - don’t rush too much. You may not even see problems on your screen that will be very apparent on your final medium if it’s different than your screen. What works on your image may not work on every possible image as well.
yes, EXR does not include color transformation, but am talking about 16bit clipping colors here…what colors look like is different conversation, am saying that there is NO values to tweak(and half float exr is much smaller than 16bit tiff, and it includs much much much grater color range)…if you colorcorrect right inside blender and output final colors from direcly blender, suree go for tiff 16bit but that’s not a good way is it?? you would first save in EXR, than open it in blender, if u want give it color managment and than save it again in other formants…BUT when u first render, tiff 16bit would not work in any CG work practicly, you have to be sure that nothing is clipping first so basically you have to applie some kind of color managment and be sure that it will affect colors heavy. good luck color correct after that
anyways, I see no point arguing here. I learned already my lesson I wont ask something like this here again. I will never ever save my original render in 16bit format tiff or png, ever after my find…I’ll delete this thread in little bit so noone wastes time reading it if its so useless to argue about.
thank you for your time, really. I know more about blender now than I knew 1 hour ago
That’s not necessarily a bad conclusion. PNG is a terrible format. And using EXR is great idea in general. There are different circumstances though. Never say never…
I like how you say you’re not giving anyone advice and then proceed to literally give advice.
Anyway, this is nothing specific to Blender, not in the slightest. 8-bit vs 16-bit, simply speaking, is just how much data is stored between 1 and 0. No matter how many numbers you put between 1 and 0 you will never have numbers beyond 1 and 0. If you want to be safe from clipping, you need to adjust the contrast/gamma of your image before saving it (unless you’re using EXR). This is the reason why raw videos from cameras with high dynamic range always have that “washed out” appearance. The camera stores the data with lower contrast so that there is more control in post-processing.
*this is with the gamma lowered before exporting from Blender.
I used to use EXR more than I do now simply because it’s generally a much bigger hassle than a properly done 16-bit PNG and sometimes EXR just won’t transfer nicely to a standard color space (probably just a “me” issue though).
But regardless, saying 16-bit PNG files are useless is completely incorrect in every way. You should not be saying such things without even knowing how bit-depth works. For beginners especially, a 16-bit PNG will be much more straight forward and easy to work with than an EXR.
Yeah. Still waiting for the deal-breaking fileformat that’s as small as a JPG but saves as much data as a 16-bit PNG…I will be waiting forever…
But generally, unless you either have a really slow PC or you’re saving a gigantic image, PNG will be fast enough. But EXR vs PNG, I just chose what works best for the specific case I’m working with.
I mean… Tiff may be more than 5 times faster. So say you are processing frames. Those tend to have some numbers… It adds up. Without any reason. I know I am not using PNG… You decide for yourself. It is slow and has no advantages. It’s not like it supports any features… No layers, no premultiplied alpha(not that it would be relevant in Blender’s context), no nothing. Transparency for web? We now have WEBP with way better compression and improving support. Why? Why should anyone use PNG?.. But it’s not my business what you use.
Just to be clear - I mostly use EXR, but for those cases that I don’t, I don’t ever find reason to chose PNG.
My main reason is just that PNGs are much more usable than TIFFs. For whatever reason, even though it’s literally 2025, a fair bit of software doesn’t support TIFF files. Plus, TIFFs only save about <4% faster than PNGs (from my tests on my hardware at least). So why hassle with using a file format that’s a whole whooping 4% faster but may not work in whatever software I need it in.
And, like I said, I do stuff on a case by case basis. Like no joke, right now I’m trying to figure out how to make a 10.8 gigapixel image to a point that I can use it. JPG can’t seem to even encode it and the PNG file is so large that no software will open it. So I’m going to have to either try another file format or something.
But, PNG is my general go-to format since every software (or at least 99.99%) accepts it but it’s not as bad of quality as a JPG.
Yeah. I can get that. I am a huge fan of doing stuff on a case by case basis. I used to experience annoying slowdowns when saving PNGs though. Other formats work fine in most cases. But sure. There are always different circumstances I guess.
It’s a very interesting problem you got there. No idea how to solve that.
The color depth of a display is at most 10bit (some maybe already at 12 bit ??)
(One may check on any display hardware comparision site and look the the selection option.)
See:
Some options to use HDR transfer functions that better match the human visual system other than a conventional gamma curve include the HLG and perceptual quantizer (PQ). HLG and PQ require a bit depth of 10-bits per sample.
and then as follow-up
Main article: High-dynamic-range video
…so one might not even see the difference ?? (Especially when using a “simple” 8-bit display ??)
On a side note: also do not forget that Photoshop’s 16bit mode is actually 15bit (plus 1 bit): 0-32768 instead of the expected 65535 value range for each channel.
The instant a full-range 16 bit file is opened in Photoshop you will lose half that information. And Photoshop will not warn you when the file is saved.
This works out from:
0-1-2-3-4-5-6-7-8-9
to
0-0-2-2-4-4-6-6-8-8
If this is important in your workflow (such as with displacement maps, etc): stop using Photoshop for your 16bit work.
JPEG files max width and height is 65,535 pixels, so that would explain it.
For WebP it’s 16383 x 16383 pixels. AVIF (without independent encoded tiling): 8,192*4,352. With tiling 65,535 by 65,535 pixels.
JPEG XL is a good candidate for your work: up to 32bit per channel, transparency, lossy and non-lossy compression, fast compression/decoding (as fast as regular JPEG), and a max pixel resolution of over a billion (2^30-1) pixels on each side. Wide colour gamut and HDR. Waaay better quality of lossy encoding compared to the old JPEG one. Animation support!
Support for arbitrary channels/components (up to 4099): CMYK, grayscale, RGB, alpha, thermal, depth, etc.
And lossless compression is much better compared to TIFF, PNG, and PSD.
Essentially a perfect working file format ready for the future, and usable now.
According to caniuse.com, “all major browser engines are working on implementing this spec.”
