HDRI Upload/download Library?

For people reading this that do not know what I am talking about.

It is the image choice applied to the chrome ball in the lighting section.

At least on my comp when I import an image I take off for a while or go to bed as it
seems to take around 4 hours to load.

But the idea is that the human eye can react to light levels far beyond what can be presented traditionally on a computer monitor.

HDRI i.e High Dynamic Range Imaging, is a method of getting around this limitation.

So as I understand it you capture multiple exposures of the same image scene and then
run these through a program that blends them in just the right way to give you a usable HDR image.

From what I have read you want at least 3 images of the same scene. And you want
everything to remain the same as possible except for your camera exposure.

And it is reccomended that you do not use the automatic exposure change settings
but to change them manually by steps of at least one EV value.

I was recently looking at some HDRI results in this particular case from within Bryce and the artist reported that the image took 5 hours to render and the result
was some minor shadow depth.

In contrast the great thing about FF it appears is that the wait time is in loading the image rather than in rendering it.

This is me just sharing what I recently learned. Please feel free to discuss, correct or add to this thread.

On the library idea since we can import our own hdri images in the format of .exr or .pfm files into FF in order to use them in our lighting choices, I just thought that if people are making their own then we could share them in the same way we are sharing
filter submissions.

And yes hdri as I understand it is combining those values into a single image.

And yes I mean't the finished images they could be in either the exr or the pfm format and be ready to import into FF for anyone with the patience.

On your last paragraph I got the first two references but the third "hdri environments for the surface lighting environment"?
Not quite sure what your referring to on that reference.

Side note; at first I had found a free utility for converting jpegs into exr's.
Which is how I experimented with it the first time.
However I just checked and sure enough CS 2 will convert jpeg and bmp formats into
pfm format which can be imported into FF.

And like I mentioned in my above post I have not done a true hdri image yet.
For my experiment I just converted a couple of normal jpegs that I had already.
And they actually do work! They imported and appear on my lighting list.
Only they are much darker than the others as true hdri's have much greater exposure
levels in them.
One of them is a picture of the head of the terminator robot after the skin has been
burned off. And it yields some interesting metal shape reflections directly onto
the image being rendered in FF.

for example this is a base image for the next example, it is just a wall with a couple of iterations of the aqua button.



This one is the same image with the 'bridge' hdri lighting choice in advanced metalizer with all color desaturated.



And this one is with the non true hdri image of T2 from my HD with all the same settings.



So as you can see it does not have to actually be an hdri image to get differing results, however actual hdri images are just that much more powerful.

So I was just wondering if anyone else is playing around with this, and/or creating
any of their own finished true hdri images.

Quote
infiniview wrote: I also did not see the option in CS 2 for conversion to exr

File > Automate > Merge to HDR
File > Save As > OpenEXR (available in 32-bit mode only)

Quote
On your last paragraph I got the first two references but the third "hdri environments for the surface lighting environment"?

sorry, that shld have been simply 'hdri environments' or 'lighting environment'. 'church', 'bridge' and those are what i was trying to say.

as for:

Quote
It looks like you had the same idea. Would they really be too large?

, i'm assuming you mean using large .exr files imported into FF for environments. be prepared for some VERY long rendering times if you start using images over about 400 x 400 pixels. as i recall, something as small as a 1024 x 1024 takes HOURS to import into FF. 200 x 400 is relatively fast, taking only minutes (depending on one's computer).

and the simplest way to get an .exr for import into FF is to simply manually change the file extension of a .jpg. simply erase the '.jpg' and replace it with '.exr', without the quotes, of course. just be prepared to have photoshop or paint shop pro choke on the now altered file shld you happen to try to load it up in one of those programs.

and, going back to your original question/request, yes, i've already suggested that it would be good to have a central repository here on the FF site, like the standard filter library, for FF environments. but, i also see no reason why some of us couldnt do this also. i'd just have to put my web server back up to do it, for example... and get ftp capability. i'm not a very good web jockey.

i wasnt either. for the library, i meant using the finished environment files. i would have just posted them in the forums, but they are well over the 250k limit.

ok so the .ffenv file is the format it becomes after being imported into FF.
So I it must be much faster to place one of these like if I got one from you and dl'ed
into mine.

Ok so here is a dumb question, an aspect I have not looked at yet, is I see people
posting filter files right here in the forums.

How is that done?

You can post a filter using the "Attach a file:" option. (It's right under the box you type in to make posts here.)
If you can't find your filter file with the "Browse..." button, then you can open the filter in FF and use the "Filter->Locate File" function (or just press Ctrl+L) and it will open a window with your filter highlighted in it's folder. (And towards the top of the window, you can see it's address.)


Edit: Oops, it looks like Crapadilla beat me to the punchline.

I think this can be organized via the wiki (we're working on it).

Well, I got it (renaming a .jpg as .exr and importing it as an HDRI) to work, but it uhh, sucked ha-ha. Can we maybe get a thread/sticky going about where to get good HDRI files, be they free or payware ? I've seen a few around online, and am interested in blowing a few bucks on some good ones in the near future, so it would be cool for all of us (I can't be the only one wanting to get some new lights eh), if we could maybe get a list going with comments from those who have used them.

jffe

Added : How the heck can I open .hdr files for lighting with FF ?

Well, posting final HDRIs are good also, since you will save a lot of time for people importing HDRI images

Quote

Kraellin wrote: and jffe, if i'm understanding your question correctly, the hdri's are the lighting environments you use in surface filters, not simple filters. it's those sphere's that move laterally, like 'forest', 'church', 'day', 'night' and so on. when you import your .exr into FF, it converts it into one of those, only based on your image and with the name you assign to it at the time of the import.

----I did that with a couple of .jpgs, and they sucked ha-ha. My question was about .hdr files, there are some free ones online, but I couldn't just change the filename to .exr and import them like with .jpgs. How can I used a .hdr file with FF like that ?

jffe

O.k. Kraellin, I guess I thought .hdr files would open/convert as easily, since they should be closer to the standard than just a .jpg is already, but then, I know basically nothing about all that. So FF team, what's up with .hdr files, and how can they be used for lighting in Filter Forge then ?

jffe

Quote
onyXMaster wrote: HDR Radiance format (a limited subset involving RGB data only) might be supported in one of the future versions of Filter Forge.

Look forward to it

Well, to start I want to note say that if you're importing correctly formatted HDRI images (latitude-longitude format), the source HDRI aspect ratio should be 2:1 (pretty obvious, 360 degrees latitude, 180 degrees longitude, that's 2:1).
And for output precision -- it varies, and varies a lot. If you have a noisy surface, have high reflection blur set or crank the reflection almost to zero, you can get away with images with a lot lower resolution than the target image (4x smaller on both axis easily, up to 10x and even more if you're willing to sacrifice some quality). On the other hand, smooth, highly reflective surfaces with reflection blur close to zero may require environment map resolution LARGER than the target image (2x at a minimum, 4x even better). I'm not going to dive into the reasons for this -- if you would like to know I recommend you to google/wikipedia for "environment mapping" or "reflection mapping".

400x200 will result in better quality distribution along the surface.
And about longitude and latitude -- IIRC in case of a planet it's 360 latitude (along the equator) and 180 longitude (along the meridians).
In any case, the 2:1 is the proper aspect ratio for input images, trust me
Also, keep in mind that since the image is a ... well, let's call it projection of a sphere surface, it's completely nonlinear. If you place a round dot with radius 10 with center at 50;50 or 150;50 on a 200x100 image, it will end up as a proper circular "light source". On the other hand, if you place it at 75, 75 for example, it will get distorted in a final lighting environment, since it will be projected on a sphere with a distortion. The safest zones to place artificial lighting are W/4;H/2 and W*3/4;H/2 coordinates if W is width and H is height (W/H should be equal to 2 as I mentioned before).

It depends on the surface shape. Imagine the front face of the sphere is what you see in a metal polished ball you keep in your hands right before your eyes (also imagine that you do not see yourself ). Now, if you place another metal ball, or any other shiny object, wouldn't the objects that are near the center affect tend to reflect in the ball more often than the ones which were at the "horizon" of the metal ball?

Try increasing surface height to create "deep canyons" -- this will lead to wider area of reflected image affecting the result (albeit fading into black due to the shading). The sharper the surface spikes, the larger area of the sphere is used, getting close to 100% as canyon walls become vertical (you won't see them when they get 100% vertical though).

P.S. HDRI import can be made faster if you're okay with the "reflection blur" giving more or less correct results only near min and max values. I'm not sure it will make in any FF release, but I can think of a dropdown box in the dialog with "Reflection Blur Quality" with Low/Med/High qualities, with Low being about 9x faster to import than High (the current default)

Quote
Now, if you place another metal ball, or any other shiny object, wouldn't the objects that are near the center affect tend to reflect in the ball more often than the ones which were at the "horizon" of the metal ball?

ok, i can see that. makes perfect sense. i'll have to look at mine again and study a bit more and redo some of my imports, since i think i had some of them backwards.

and, i'm really not worried about faster with the imports. i dont do that many and i can always just set it up to run before i go to bed. i've mostly only done about a max of 400 x 400 so far and that doesnt take more than about 20 or 30 minutes, if i remember correctly. and since most of the images i'm currently doing are only at the 600 x 600 or so size, it hasnt been that important to have a higher resolution environment. but, i may try something like a 1200 x 600 import just to test it.

thanks, onyx, this has been helpful