Hd For Indies

[Dave Cooper]

[Gavin Greenwalt]

Had some time to blow during a render to do a quick slapcomp:

[David Battistella]

I made a 'discovery' while doing a sky replacement on the magnum:



Compression tiles. Yay!

If you look closely at the hood of the car in the close up I did you might see a what looks like a dead pixel.

David

[Jaime Vallés]

If you look closely at the hood of the car in the close up I did you might see a what looks like a dead pixel.

David

Where?

[Steve Sherrick]

I've been hesitant to point some things out about the uncompressed TIFFs but since the can of worms has been opened...

I've seen dead spots (clipped areas perhaps) and noticeable compression. Perhpas in motion, these don't seems as noticeable. I can't argue with the range I've been able to play around with when doing some grades, but looking very carefully at the images, I've seen some things I have questions about.

I'm actually hoping that I'm somehow not looking at the absolute RAW file somehow, maybe there was some artifacting introduced by TIFF?

I was working with the images in Photoshop for what it's worth.

Steve

[Gavin Greenwalt]

Tifs don't introduce artifacts.

[Steve Sherrick]

Uncompressed TIFFs don't as far as I know, which is what these were so yeah, i would tend to agree. But I am clearly seeing compression and white blocks in the images.

Just trying to understand what I am seeing. I'll put together some pics to point out what I'm referring to.

Steve

[David Battistella]

Where?

http://www.reduser.net/forum/showthr...?t=4330&page=4

it's about 45degrees northeast of the inside headlamp on the hood. I can't tell if it's a pixel or just a spec of dirt on the hood.

David

[I Bloom]

I made a 'discovery' while doing a sky replacement on the magnum:



Compression tiles. Yay!

That's interesting oneguy. I guess that's an example of the compression algorithm choosing to drop higher frequency data in that one area presumably because of that section of the car and sky having really low detail. It doesn't seem incredibly troubling that it exists that way, but it might have consequences especially when heavily grading moving images. It makes me think that the RAW port might have some real applications when you know the image has to be pushed hard in post.

I'm away from my workstation so I have some questions:

1. How large is each compression tile?
2. Once you have some actual Redcode footage: How far can we push the image in grading before we see some noticable dancing tiles?

By the way, Concrete and Offhollywood. These images look really awesome. It's amazing to think that most of what we've been talking about for the past year has all been "in theory". I like talking about the real thing much better.
Mark great call shooting all white cars.

IBloom

[Ryan Sims]

Complaining about compression and then posting a JPEG showing the problem area may not be the best solution. Here's a PNG at 100% zoom of the same area pulled from the original 48MB TIFF. The PNG is 1/3 the file size of the this same pictures' TIFF. Does a PNG introduce any new artifacts?



When looking at the raw file at 100% zoom with no adjustments, you can't see any of these compression artifacts. You have to push it very hard to see any tiling. Graeme has pixies working on it, but let's face it, it is not magic. Real compression has to take place somewhere. Considering the 48MB size of a single 4K TIFF frame, the REDCODE RAW data takes up only something close to 1MB for that same single frame. I just hope we will be able to post single frames in the REDCODE RAW format in the future and everyone be able to view them with a free version of REDCINE.

Here's a link to the TIFF: http://www.mediafire.com/?2xbdneo23nz