10+ Free HDRIs - 16000x8000, up to 24 EVs
polycounter lvl 5
Update: I'm now publishing a free 16K HDRI every single Friday - see the latest posts for images or just head here for them all 
Original post below:
For the past few months I've been working on a dedicated website for all my HDR panoramas called HDRI Haven. In preparation for its launch next week, I've got a little gift for you
Original post below:
For the past few months I've been working on a dedicated website for all my HDR panoramas called HDRI Haven. In preparation for its launch next week, I've got a little gift for you
Usually I only give away small HDRIs (2048 pixels wide or less), but this time you can have five giant HDRIs (16384 x 8192 in resolution). All of them contain the highest dynamic range possible, meaning they'll give you perfectly realistic lighting right off the bat without requiring any complicated node setups.
And some test renders done in Blender to show the lighting that they produce (the HDRIs are the only light source, the node setup is simple):
PS: Sorry for the email form thing, Gumroad is the only decent platform I've found that can host these giant files for free without any kind of annoying speed limiter to get you to pay for it. And no, Dropbox and Drive don't work reliably as hosting platforms, not at this level.
PS: Sorry for the email form thing, Gumroad is the only decent platform I've found that can host these giant files for free without any kind of annoying speed limiter to get you to pay for it. And no, Dropbox and Drive don't work reliably as hosting platforms, not at this level.
Replies
I just have a canon 600d and a 10-18mm lens. The trick is using strong ND filters and some custom software to get the super high dynamic range.
The zip file should be 381 MB, and the 16k .hdr file should be 370 MB. If the file sizes don't match, maybe the download timed out?
mpumalanga_veld_16k.hdr: F03121A4CC66493F0F55F9AECCAECA18
I did download it from gumroad myself, but someone else mentioned an issue with the same file a while ago so I'd just like to double-check it's an incomplete download problem.
I'm curious what tripod/mounting setup you use to get the 360 coverage and also what custom software you are referring to? I've been interested in getting into this for quite some time now but haven't gotten around to it. This made me all the more eager
About the software, I was struggling to get any program to correctly merge the exposure brackets in all cases - some would introduce lots of noise and hot pixels, others would get the brightness all wrong, and others would create artifacts around bright light sources. Having 24 EVs of dynamic range in some HDRIs compounded this problem, so I ended up using Blender's compositor to manually merge the exposures and wrote some tools to automate this process. "Custom software" is perhaps too fancy a term, more like "some scripted hacks upon existing software to get it to do things it was never meant for"
Check it out: https://hdrihaven.com/
How many sets do you need to take with that 10mm? I'm using the Samyang 7.5mm fisheye, 6 around, 1 up, two down. I get 12x6k files out of a 16mp sensor, so probably not good enough for you if you're looking for 16x8 (though a 24MP Sony E mount body would probably do around ~16x8).
I'm also curious to know how you're measuring the dynamic range of the files. Lys can read the unique value range and we've done some math to figure out that our Panos get up to about 25 stops of DR, that's with EV sets of 7x2EV or 5x3EV (gives about the same range). So I'm curious to know if that's 24 stops of DR, or if your exposure range is 24 stops of EV (ie, 14x2EV, which would be much more than 24 stops of DR).
The exposure range is 24 actual stops. I've had no luck making any sort of accurate measurements, but I'm no mathematician. All software I tried that measures dynamic range gives a different result, so it's not a good metric to compare with other people's HDRIs. Better to use the actual number of stops taken, so 5 shots spaced by 3 EVs (range from 1/4000th to 1 second) is 12 EVs.
If you're taking 24 EV, so about 10x3EV or whatever, you're most likely capturing (or have the potential to capture) well above 24 stops of dynamic range. A single capture in raw at base ISO is about 10-14 stops of DR for reference. I don't really have better advice for measuring though, I only barely understand the math myself. =P
In all likelihood you're capturing more data than you really need to, which will result in a lot of essentially black frames, and may be why you had problems with various HDR merging software. In my experience, most scenes don't need more than 12EV range, as lot of scenes have significantly less dynamic range than that. The primary exception being the sun if it's not occluded by skies, then 12EV starts to clip unless you underexpose a bit to retain the highlights. Exposing for the highlights rather than the shadows helps in general, and if you're getting bad results with 12EV (like you show on your site), you may have the exposure range set up poorly.
For example:
This is 7x2EV, and as you can see, as long as you're maximizing the exposure values, it captures essentially the full range of light, though of course the sun is brighter than it appears there so it's clipping slightly.
Anyway, you clearly know what you're doing, and if you don't mind the workflow your using there is no reason to change it, but you could probably speed up your capture and processing workflow quite a bit.
You could capture the full dynamic range of the sun with even 4 EVs, it would just mean the rest of your image would be almost solid black and noisy as crap. Maybe I exaggerate. Likewise you could probably shoot 30 EVs and still clip a light bulb if you really tried (high ISO, start with a very long exposure). This is why even bothering to tell people what EVs of dynamic range you shot is pretty silly in the first place. It just seems to be the standard so people expect to see it. I think it would be fine just stating that an HDRI is unclipped (as long as you don't lie about it like some people do).
Getting the 24 EVs isn't too inconvenient, and I like to make sure there's plenty of detail in the shadows in case someone wants to use it as a backplate.
I have more thoughts on dynamic range but I need to research and verify some assumptions before I go too far here. I think it would be great to have a standardized method for measuring and describing the DR of an HDR file.
Get them here.
This one's my favourite
You can find them at the top here marked with a green "CC0" flag: https://hdrihaven.com/hdris.php?sort=popularity&npp=100
So far there are 6 of them, but here are 3 of my favourites:
Here are weeks 9, 10 and 11:
Download
Download
Download