Welcome to our community

Be a part of something great, join today!

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 :)

Color space...

And in Epic, for those that want losslessly compressed (same output as uncompressed, just smaller file size) they'll have that option too.

Graeme
 
Graeme Nattress said:
And in Epic, for those that want losslessly compressed (same output as uncompressed, just smaller file size) they'll have that option too.
Click to expand...

Yee ha! There are times when I've truly wished for that—usually for exterior stills, where a billion or so leaves stress any codec pretty radically. Graeme, what will the lossless Redcode record to?
 
so just so i'm understanding this correctly. Because RAW goes by shear resolution and amount of pixels on its information instead of subsampling to gain its color information, the amount of "color" you can push around in post is simply left up to RAW interpretation software, codec and medium you use for the color manipulation, and actual ability of the camera. (sensor, lens, etc.)

and in combination with the DR of the sensor and how the acquisition codec resolves that (example: compressed lossless in epic) what is the the measurement for how much color we get out of the picture?

isn't 4:4:4 determined to be the most amount of color video can have under broadcast standards? if 4:4:4, 4:2:2, 4:2:0, don't really apply to RAW, how do we gauge? just pixel count?

simply saying RAW is RAW, doesn't explain enough to me.
 
4:4:4 etc are video terms and relate to YUV compression by chroma subsampling - basically storing a lower resolution chroma compared to luma. So, if you don't use that kind of compression, those terms are meaningless.

The thing to remember, is that even VHS can be stored as 4:4:4 so alone it means nothing with respect to any kind of quality.

For RAW, you have three things - the number of pixels, the pattern of those pixels if the sensor uses a colour filter array (like the Bayer pattern we use) and the quality of any compression used (REDCODE is 12bit wavelet based compression).

Graeme
 
Yep, color subsampling is just a method for "sampling" luminance and chroma data from the source. Therefore, 4:2:2 subsampled video could be better than 4:4:4 subsampled video depending on which source the video was subsampled from.

Bottom line: how much data is available on the source is the most important factor.
 
4:4:4 means better color control?

4:4:4 means better color control?

Aaron Newsome said:
I think this thread would have been a lot shorter if the response to the 4:2:0 questions would have been, that 4:2:0 describes color subsampling, a fitting of data really. RED doesn't subsample, it record the raw data. All the pixels, all the values.
Click to expand...

Well it is enlightning to know , coz i only knew that 4:4:4 is the colorspace best for chroma keys and also for better coloring. IS 5d also shooting in RAW video.

From all i learned from this thread is that Red records RAW which simply means the best image control possible in post. am i right>
 
The Canon 5D2 shoots what is practically the opposite of RAW video - highly compressed, poorly downsampled, line skipped, over-compressed h.262 Y'CbCr 4:2:0 not RAW.

RAW does indeed give best image control in post.

As noted above 4:4:4 is not a colour space.

Graeme
 
RAW / 4:4:4 etc

RAW / 4:4:4 etc

What Graeme says is the correct technical explanation.

I'll offer the below as a "marketing explanation " i.e. a reasonable way of "explaining" this to producers in words they "understand"

" RED shoots RAW, which is equal to or superior than 4:4:4 "
 
Hey,

Graeme and Stuart, I'm going to ask the ? I posed in another thread here. Does Red apply a gamma curve to the Bayer sensor values before compressing to 12-bit .R3D? Or is no gamma curve applied (I believe the correct term for that would be linear light)?

I'm asking b/c, FWIU, many cameras add a gamma curve before compression and claim this helps reduce noise in compression. But it seems that it would also go against the RAW ideology, so to speak.

So I'm curious what Red does. Mostly b/c I'm wondering if when I choose Linear Light in RedCine-X if a gamma curve added during compression is removed in RedCine-X.

Or was no gamma curve added during compression, in which case Linear Light would straight from the sensor vlaues. (Well, less compression losses, demosaicing and color space adjustments, but no gamma adjustment.)

Thanks much.
 
All colorimetry must be done in linear light, so the R3D codec is designed so that it's output is always linear light. Any gamma adjustment is therefore only done on linear light data after colorimetry. That also makes it easy for us to offer the EXR format, which is inherently linear light.

Graeme
 
Nook Kim said:
Yep, color subsampling is just a method for "sampling" luminance and chroma data from the source. Therefore, 4:2:2 subsampled video could be better than 4:4:4 subsampled video depending on which source the video was subsampled from.

Bottom line: how much data is available on the source is the most important factor.
Click to expand...

A is it really a fact that 4:2:2 subsampling could be better than 4:4:4??
 
Sure it can - 4:2:2, 4:4:4, 4:2:0 tell you NOTHING about the quality of the video, about compression artifacts, or sensor aliasing, or source video resolution, or dynamic range. They only tell you if in the codec that chroma was sub-sampled with respect to luma. Or put it another way, does 4:4:4 SD, say, look better than 4:2:0 8k Scanned IMAX? :-)

Graeme
 
Ok this is really very weired, coz according to my limited knowledge i thought red camera was using 4:4:4 which to a novice like me means better color correction ability, if it is not using that , then does it mean that RedRAW format is something lossy wavelet codec . And i am not getting what i think i am getting by using red camera. damn! can someone make a simple explanation please, it would be very hurting to know i dumped my 5d for red camera rentals.
 
Graeme Nattress said:
The Canon 5D2 shoots what is practically the opposite of RAW video - highly compressed, poorly downsampled, line skipped, over-compressed h.262 Y'CbCr 4:2:0 not RAW.

RAW does indeed give best image control in post.

As noted above 4:4:4 is not a colour space.

Graeme
Click to expand...

Ok thats what i needed to hear and thats what matters, that without going into engineers course in video compressions i can be sure that by using red ill be getting the best image control and best possibilities in post coloring too. Thank you i am relieved .
 
Sometimes it's hard to answer technical questions without getting overly technical. On the other side, technical terms are thrown around all over the place and used in ways that defy their meanings. One of the things we try to do at RED, other than just offer superb quality image quality from our cameras, is explain terminology and technology on this forum, and answer questions about cameras, with the answers coming from the people who know, because they are intimate with the design of the camera, rather than answers coming second hand from marketing.

Graeme
 
Graeme Nattress said:
Sometimes it's hard to answer technical questions without getting overly technical. On the other side, technical terms are thrown around all over the place and used in ways that defy their meanings. One of the things we try to do at RED, other than just offer superb quality image quality from our cameras, is explain terminology and technology on this forum, and answer questions about cameras, with the answers coming from the people who know, because they are intimate with the design of the camera, rather than answers coming second hand from marketing.

Graeme
Click to expand...

So true and let me tell you after reading thousands of forums atleast i can say i am convinced by your words because you guys are THE people dealing with this stuff. Haha but its funny when i learned that what i knew earlier about subsampling was infact not true..Thanks
 
mansh said:
Ok this is really very weired, coz according to my limited knowledge i thought red camera was using 4:4:4 which to a novice like me means better color correction ability, if it is not using that , then does it mean that RedRAW format is something lossy wavelet codec . And i am not getting what i think i am getting by using red camera. damn! can someone make a simple explanation please, it would be very hurting to know i dumped my 5d for red camera rentals.
Click to expand...

You're focusing on all the wrong things. 4:4:4 is irrelevant here really. What is most relevant about storing raw images, is the pixel values from the sensor are read from the sensor and stored in the image file with the absolute least amount of PROCESSING.

PROCESSING DATA = throwing data away.

RED does not throw data away. It stores all the original values read from the sensor so that you have the full range of information available in post.

A 5D on the other hand does just the opposite. A 5D does a whole lot of processing of sensor data, well before it is passed to the encoder and then stored as H.264. In post, you don't have the full value of data as read from the sensor. You have a thinly sliced, processed and optimized version of the sensor data as it was stored in the H.264.

This is why when you try and grade highly processed footage, it falls apart very quickly. This is not a myth. Try it yourself and you'll see.

If you try and push highly compressed images, if you try and recover details in the shadows and so on, it just doesn't work. The data was thrown away as part of the processing done by your camera.
 
and just last question coz i asked a lot already , on wikipedia digital cinematogrphy page there is a list in which it states red raw as LOSSY ?

DCODE RAW 12 bits 4520x2540, 4480×1920, 4096×2304 Raw Bayer 224-336 Mb/s 1.6-2.5 GB/min. No Wavelet (lossy)
ARRIRAW 12 bits 2880×2160 Raw Bayer ~ 5.6 Gb/s 42 GB/min. No Uncompressed
"DALSA" RAW 16 bits 4096×2048 Raw Bayer ~ 3.2 Gb/s No Uncompressed

http://en.wikipedia.org/wiki/Digital_cinematography
 
Redcode is lossy or lossless, depends on the Redcode rate. At its highest rate it is mathematically lossless.
 
You must log in or register to reply here.
Back
Top