Welcome to our community

Be a part of something great, join today!

Register Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Have the compression algorithms improved on Dragon?

My understanding is that because of the new color science, etc. compression artifacts are reduced. But I'm sure some code had to be rewritten in order to accommodate the substantially larger data pipeline required for 6k(?).
 
Yeh it has been stated a few times that the increased signal to noise ratio and lower noise floor helps the compression a lot. I'm truly shocked at the detail in the 17:1 shots, especially the last shot with the water breaking in black and white. Just incredible. I'm pumped to see some R3D's to see if its as good as it looks.

Its weird I noticed a lot more noise in the stills that Mark posted compared to the footage. It make me wonder if Mark ran the footage through some noise reduction or if the noise being so small just got lost in the h.264 compression, Or was it the Jpeg compression from the photos that introduced noise that wasn't actually there...could be a lot of things.

All around I can't wait to get my hands on one
 
Joseph Coleman said:
Yeh it has been stated a few times that the increased signal to noise ratio and lower noise floor helps the compression a lot. I'm truly shocked at the detail in the 17:1 shots, especially the last shot with the water breaking in black and white. Just incredible. I'm pumped to see some R3D's to see if its as good as it looks.

Its weird I noticed a lot more noise in the stills that Mark posted compared to the footage. It make me wonder if Mark ran the footage through some noise reduction or if the noise being so small just got lost in the h.264 compression, Or was it the Jpeg compression from the photos that introduced noise that wasn't actually there...could be a lot of things.

All around I can't wait to get my hands on one
Click to expand...

When we get our hands on TIFFs we'll know better.
 
Robert Ruffo New said:
Looking at Mark's tests, it would seem that they have.
Click to expand...

Doubtful, the compression per pixel stays the same but there are more pixels so it's a higher bitrate for the same ratio and you are still watching it on the same size 1080p screen. 6k 17:1 is still ~40MBps which is about the same as 4k 8:1 which looks more than fine for 1080p delivery... Now if you were to look pixel for pixel you might see a different story, but 1:1 pixel magnification is a higher magnification for 6k than for 4k for the same delivery size...

The algorithm seems to be the same but likely does compress a bit more efficiently due to lower noise levels.
 
Just guessing that the R3D algorithm may have been tweaked as no software will run Dragon footage without a rewrite or a plugin?
 
I wouldn't be surprised if they tweaked things a touch, but as I understand it is at least mostly because the bigger the image the more efficient their math gets. It makes perfect sense. So file size is not linear with respect to resolution.
 
RED/Jim has said that because the larger Signal to Noise ratio and improved noise floor that 12:1 is equivalent to 6:1 or 8:1 or something like that...

... in any case, yes, it has been improved. I mean, shit, 20mp instead of 14mp is a substantial difference (almost 33%).
 
Joseph Coleman said:
Its weird I noticed a lot more noise in the stills that Mark posted compared to the footage. It make me wonder if Mark ran the footage through some noise reduction or if the noise being so small just got lost in the h.264 compression, Or was it the Jpeg compression from the photos that introduced noise that wasn't actually there...could be a lot of things.

All around I can't wait to get my hands on one
Click to expand...

the stills are always noisier. Do a test in rcx. I think it's because r3d is compression designed for motion and not stills. The Mx at 800 in tungsten produces noisy stills but when it moves it looks good.
 
Eric Haase said:
the stills are always noisier. Do a test in rcx. I think it's because r3d is compression designed for motion and not stills. The Mx at 800 in tungsten produces noisy stills but when it moves it looks good.
Click to expand...


I'd like to see more elaboration on this point. What about all the fashion shoots done with Red?

I don't buy the "stills are noisier" theory… unless they truly are noisier but it's so negligible it's not even worth mentioning ?
 
zak forrest said:
I'd like to see more elaboration on this point. What about all the fashion shoots done with Red?

I don't buy the "stills are noisier" theory… unless they truly are noisier but it's so negligible it's not even worth mentioning ?
Click to expand...

You can see it for yourself. Shoot some footage at ISO 800 - preferably under tungsten light in a studio and look at stills. There is noise but it appears minimized when you play. There isn't actually less noise- it just appears that way. Part of it has to do with the fact that the Epic is noisy in the 0-15% IRE range. Most of the time this tone occurs in the background (like a painted wall) and rarely on a subject's face. So during motion you're attention is mostly directed to the moving object which has less noise than the background. In a still you have more time to look at everything in the frame. Just my theory from observation but try it for yourself.

And regarding the fashion shoots... I would suspect that they rate the camera lower in a controlled studio environment like at 320 and use plenty of light and fill the histogram. Also, they can reduce the noise in the stills in post. The stills are usually retouched a lot.
 
The cleaner the sensor, the easier compression is. Dragon is cleaner.

Jim
 
Ahh, so you are saying we are paying $9500 to have our Epics cleaned? ;-)
 
On a serious note, thanks for putting the best camera in the world in my hands, TWICE in 2 years. Forever greatfull
 
Jannard said:
The cleaner the sensor, the easier compression is. Dragon is cleaner.

Jim
Click to expand...

Wonderful news!
Is there going to be REDCODE 1:2.5 option in Dragon's firmware? What is happening with availability of RED SSD TURBO MAG 512 GB and is there any faster solution on the way? Thanks in advance!
 
Working on it...

Jim

Filip Orlandic said:
Wonderful news!
Is there going to be REDCODE 1:2.5 option in Dragon's firmware? What is happening with availability of RED SSD TURBO MAG 512 GB and is there any faster solution on the way? Thanks in advance!
Click to expand...
 
Joseph Coleman said:
Its weird I noticed a lot more noise in the stills that Mark posted compared to the footage. It make me wonder if Mark ran the footage through some noise reduction or if the noise being so small just got lost in the h.264 compression, Or was it the Jpeg compression from the photos that introduced noise that wasn't actually there...could be a lot of things.
Click to expand...

You are comparing temporally compressed h264 to JPG stills. The h264 is compressed to on average dedicate about 50KB/frame, whereas the JPG is compressed at about 500KB/frame. The first thing you lose when applying lossy compression is usually the appearance of noise, because it's randomness spatially and temporally make it the hardest part of an image/video to compress.

As for R3D stills and motion having noise that looks different I would guess that noise is a bit harder to see when its moving as it may tend to average out a bit temporally to our eyes/brain. Just a guess, though I imagine there are a number of things that may allow us to see noise better when it is not "moving".
 
will the new Sata Express standard be usable/adaptable for RED Dragon/SSD's since it allows over 4 times the speed of Sata III
 
You must log in or register to reply here.
Back
Top