Lucas Digital Releases OpenEXR Format

frankie writes "Although George Lucas may have gone over to the dark side, at least some of his staff prefer Freedom and light. ILM has released OpenEXR, a graphics file format and related utilities, under a BSD-style license. Among other things, it supports the same 16 bit format used by Nvidia CG and the Geforce FX. OpenEXR runs on Linux, Jaguar, and Irix; other platforms are likely to work with a little help from the community."

Comments

It's cool that they have the file format

[exp(pi*sqrt(163))]

But have you watched any movies with ILM effects lately? The dynamic range sucks! Episode II was basically characters jumping between matte paintings and each painting looked like it had been painted with an 8 bit paint package. Unless you actually bother to collect data on set that is high dynamic range having the file format is as good as useless.

Not off topic.

[Anonvmous Coward]

"Out of curiosity, has anybody used HDRI images for textures? I'm curious if the floating point data makes a difference. I could see it being particularly useful for the diffuse and lumination channels. What about color?"

Okay, somebody modded me down as 'Off-Topic'. I'm just going to assume he/she/but probably he didn't understand what I was talking about here.

OpenEXR is a format for High Dynamic Range Imagery. What this essentially means is that instead of describing a pixel by having 3 channels @ 8-bits per channel (which has a maximum value of 255), you get a floating point 16-bit value per channel which is a measure of intensity. The result? Instead of having just color data there, you have color data & intensity data. The sky's blue, right? If you take a 24-bit picture of the sky, you get blue pixels. Is that enough data? No. Try looking up at the sky without squinting your eyes. Can't do it, can ya? The sky is *very* bright. With the HDRI format, you can store that luminosity as well as the color. That's why they use it for global illumination. You're capturing light sources, intensities, and color at the same time.

Thing is though, a floating point format has uses in other areas of 3D such as texture mapping. It means you can create/capture textures that deal in intensity as well (just like real life), thus you get a much more realistic response from lights in the scene.

I have no idea if I'm making any sense here or not, but the main point I'm trying to make here is that I am nowhere near off-topic. That's the reaason this format is interesting. It's not another .PNG or .JPG format, it's a more accurate way of storing information about light, and us people that work in 3D have a lot to be excited about. Since it's just recently become involved in the major 3D Apps out there, the capabilities of it are still in their infancy and I'm curious what people have discovered about it.

Tiling is irrelevant

[Namarrgon]

Thousands of layers? The most complex composition I've seen personally was the Swordfish Ventura Bank explosion, and that required somewhat over 500 layers (at 4K). Definitely qualifies as heavy duty, but still a far cry from "thousands" of layers. "Often" would be less than 100 layers, in my experience.

Anyway, tiling as you describe is rarely used in motion picture image processing work, regardless of the number of layers. Breaking down a large (4000x3000 or larger) image does improve memory usage (sometimes at a cost in efficiency for certain algorithms), but when this is done, it's usually broken into scanlines or groups of scanlines, not square tiles. This works just as well and fits better with how images are processed, stored, displayed etc. The number of layers to be composited does not affect this at all.

DPX and Cineon do not support tiled image packing. TIFF does, but I've never seen a post-production app actually output a tiled image - it just complicates things unnecessarily.

And it's rarely necessary to re-read an entire image if you just want a subrectangle of it - many formats make it relatively easy to read a limited region. Compression can complicate things, but you can usually limit your reading to just the scanlines involved.