We had the good fortune of having Ted Schilowitz from Red in Singapore to give us a demonstration of the new Red Epic, some interesting photos from Red’s past (including their first prototype which was enclosed in a applebox like wooden crate), a 4K projection of their Mysterium X showreel and their HDR motion picture implementation.
Even though the screening facilities were not geared up for proper 2D 4K projection, the richness of the image completely blows you away. The cinema we were using was the only one equipped with a 4K projector but its primary use as a stereoscopic 3D theatre meant that it was using a silver screen, not ideal for 2D images as it does not give a proper rendition of the black levels.
The opening shot of the new MysteriumX sensor showreel was dim, shot at ASA2000.
The image had little visible noise, looked like ASA 640 at worst.
Leo DiCaprio in the shot is sitting on the floor lighting a cigar. When he lights the match, THE FLAME IS ENOUGH TO ILLUMINATE HIS FACE!
See that video here. No grading, encoded in Prores, so if you don’t have the codec installed, don’t bother downloading it, you won’t be able to view it. This pales in comparison to the 4K version, but it’s the only way you can get a meagre approximation of the dynamic range and noise levels of the image. The picture has tonal range and subtlety that even ProRes struggles to do justice to.
Cynics may say the lighting was augmented, but sure didn’t look like it, the flickering and light fluctuations were true to physics.
Then, the demo of HDR in Motion. Only ever been done with stills before this. The implementation is also seamless. Still frame of the shots we saw in motion here.
Once I saw the specs stated that overcranking was possible without downrezing, the lightbulb in my head went off. For the longest time I was trying to achieve HDR in post by getting Ian from Widescreen Media to shoot 2 identical shots at different exposures with the beam splitter rigs he uses for stereoscopic work, then trying to blend these in post. Red’s implementation is far superior and seamless.
HDR is done by overcranking and exposing alternate frames at the 2 different exposure settings, and the frames are linked by metadata. I believe their term for it was conjoined frames. The cost is only increased filesize (by implication, higher throughput requirements?).
If I’m not mistaken, the filesize increase is not doubled, so there is likely to be some form of lossless/highpass/lowpass compression to save space/datarate.
eg you’re taking the over-exposed frame for shadow detail, you probably don’t need to read/encode highlights. All conjecture on my part so it would be interesting to know how it’s done.
There is an easy mode that tonemaps/blends the 2 images, or a more advanced mode that gives you more control over the blending settings, with adjustable response curves.
With a theoretical 13 stop latitude, and assuming a 4 stop overlap between the exposures, a dynamic range of 22 stops is possible. The shots on the reel looked like they overlapped conservatively by 7-9 stops (15-17 stop latitude), so I’m wondering how far it can be pushed. Bear in mind I watched it only once, and was busy picking my jaw up off the floor so my dynamic range estimations should be taken with a shovel of salt.
All in all mind blowing stuff. This changes everything, again. But without the snarky ads and hipper than thou attitude. Just guys who love their cameras, giving you the best camera technologically possible, NOW and CHEAP.
0 responses so far ↓
There are no comments yet...Kick things off by filling out the form below.
You must log in to post a comment.