2k workflow

Emanuel

Emanuel

Cross-Examiner
In the section of the Format Options regarding the Lens Format board, there is mention to 2K windowed and 2k scaled -- what does it mean?

What's the best? Or better yet, which are the differences between them?

Considering my interest in 2.35:1 format, avoiding the anamorphic option (or maybe not?), 2k will it be the realistic one to that purpose?

Specially considering the limitations @post?

I'm afraid that 4k by now it's just a dream for the future -- or am I wrong?
 
2K scaled v's 2K windowed

2K scaled v's 2K windowed

Emanuel,

Scaling means that 4K's worth of sensor data is reduced (by scaling) to 2K resolution. So that 4096 x 2304 pixels becomes 2048 x 1152 pixels. The advantage of this technique is that the 2K is oversampled in H and V, which means the quality of the 2K image is extremely high (maybe you have shot HD and downressed that to SD and compared it to video shot directly in SD. SD material originally shot in HD looks better because it is oversampled relative to the SD footage) The disadvantage (if any) is that the maximum resolution & frame rate you can operate at in this mode is 2K at 60fps.

Windowing is a technique only available with a CMOS sensor, where you can choose not to read the entire sensor area, just an area equivalent to 2048 x 1152 pixels. The advantage of this technique is that the maximum resolution / frame rate you can operate at in this mode is 2K at 120fps. The only issue to consider is that the video is not now oversampled. And because you are only using the center of the sensor when in this mode, your 35mm coverage lens will take on the characteristics of a Super 16mm lens - that is it will become more telephoto and exhibit a deeper depth of focus.

Also see http://www.red.com/product/format-options/
 
2.35:1 Imaging

2.35:1 Imaging

If you are familiar with the concept of Super35mm, a single master image is shot and used to create 2.35:1, 1.85:1, 1.78:1 (16:9 HDTV) etc. Several motion pictures have been shot that way including Terminator 2 and Titanic.

Consider the RED camera to be a digital version of a Super 35mm camera, and follow the same concepts. There is no need for anamorphic lenses and so no issues with slightly differing H and V resolution, focus or film grain / digital compression footprint as a result.

This is why RED's pixel count is greater than 4096 pixels wide - that is for 35mm coverage, 4520 is Super35mm coverage - 4520 x 2540 pixels. Then cropping that for a 2.35:1 ratio would give 4520 x 1920 pixels or 2260 x 960 pixels when scaled to a "2K" format image.

Post production systems are going to have to do a little catching up here as the maximum typical Widescreen resolution is 1920 x 816 pixels. RED's 2K Widescreen has almost 20% more resolution, so its worth adjusting NLE solutions and printers to exploit that - especially for 35mm release; but we can scale again down to the smaller 1920 x 816 pixels if necessary.
 
Stuart,

Thank you very much for the excellent enlightenment -- very hopeful!

Stuart English said:
This is why RED's pixel count is greater than 4096 pixels wide - that is for 35mm coverage, 4520 is Super35mm coverage - 4520 x 2540 pixels. Then cropping that for a 2.35:1 ratio would give 4520 x 1920 pixels or 2260 x 960 pixels when scaled to a "2K" format image.
Click to expand...
But if 2K@RED it is 2048 x 1152 we'll have 2048 x 871 (2.35:1 ratio) -- right?

Otherwise, by now there isn't any affordable editing system with 2260 x 960 support or am I wrong?

Stuart English said:
Post production systems are going to have to do a little catching up here as the maximum typical Widescreen resolution is 1920 x 816 pixels. RED's 2K Widescreen has almost 20% more resolution, so its worth adjusting NLE solutions and printers to exploit that - especially for 35mm release; but we can scale again down to the smaller 1920 x 816 pixels if necessary.
Click to expand...
But it isn't necessary if the Apple has been announcing 2K@FCP and the same it happens regarding the Premiere Pro too -- or isn't it?

And if to profit with the Super35mm coverage we're shooting via RED-RAID 4520 x 1920 (2.35:1 format) is there the possibility to scale down to 2048 x 871 just in order to handle@post production -- while 4K it isn't available as affordable solution? Considering those two NLEs (FCP & Premiere Pro) as 2K suitable?

And always considering that we want the best as we can get for 35mm film-out purposes from the digital master (online).

Why shooting@4k if it's possible scale down from the 4520 x 2540 full sensor (1920@2.35:1) to 2048 x 871 recording? Maybe to the future. Specially when 4k editing/post production will be as usual as real. If it's viable to back-up native 4520 x 1920 (for stock footage/library aim, for instance) why the recording option will be inferior?

And in this case -- if possible, where's this (hypothetical) operation (4520 x 1920 -> 2048 x 871) takes place? Inside RED-RAID (is it possible? or from there?) or just@NLE? I just can't understand how would it be possible from 4K or 4K+ if the NLE is limited to 2K...or this limitation it occurs only to editing and not for a scale down operation? Is that so?

:dankk2: again and now in advance!

Cheers
Emanuel
 
Last edited:
Here's a thought..
As opposed to cropping the image to obtain the 2.35 ratio, has it occurred to use the same system as the viper? Whereby a 2.35 ratio section of the sensor is output , via a pixel aspect of 1.33, to a full 16x9 image. So in effect you would have a 2.35 image queezed into a 2048 x1152.
Now while this doesn't seem particularly useful at first hand, remember that you're now using the entire bandwidth of your REDcodec to encode the 2.35 image, rather than wasting 30% of it on material that's going to be cropped.
Remember also, that no matter what your eventual output media is going to be , HD, SD or film-out, your post is being done at higher resolution, and the eventual downrez to your delivery format will benefit from the sub-pixel detail aquired by working at the 30% expanded resolution.

J
 
Compression

Compression

John, assuming that your codec locks you into a traditional 16:9 HDTV frame size, then an anamorphic scaling as you describe is a very reasonable approach, although you now have some additional (potential) scaling artifacts to deal with. An alternative approach is to design your codec to accept a 2.35:1 aspect ratio input - and just encode that data. Or go ahead and encode the whole 16:9 aspect original image and then crop a 2.35:1 and a 1.85:1 and a 1.78:1 (16:9 HDTV) and a 4:3 (SDTV) out of that. Lots of choices.
 
Stuart English said:
An alternative approach is to design your codec to accept a 2.35:1 aspect ratio input - and just encode that data.
Click to expand...


Excellent, all I was basically after was the option to use the full bandwidth of the codec on the final 2.35 image. this will accomplish that nicely.

J
 
As I already posted:

Gordon Prince said:
It seems it isn't 2.35 but 2.39 and 2048 x 858 pixels (2.39) as the DCI specs for 2K digital cinema.
http://www.dvinfo.net/conf/showthread.php?t=73528

However, a dude that worked on Superman posted 1920 x 818 (2.35).
http://www.dvxuser.com/V6/showpost.p...2&postcount=12

But if 2.35 is an older format than 2.39 why should Superman be 2.35 instead 2.39?
Click to expand...
It seems quite strange that someone like Stuart English is speaking about 2.35 when the correct ratio is 2.39 if it is right.
 
Up until 1969, 2.35:1 was the 35mm anamorphic ratio used by CinemaScope and Panavision. They changed the standard to 2.39:1 in 1970, but people still refer to it as 2.35 simply due to convention. Sometimes it's even rounded up to 2.40. So you can't really get down on someone who is just doing things the way the industry does.

Furthermore, how you shoot your film and what aspect ratio you choose is entirely up to you. Film is art! Who's to say that 1.85 and 2.35 are the only "appropriate" ratios for shooting? Personally, I think there are other outstanding ratios that are not "standard" and even though they aren't utilized much (if at all), I plan to shoot in some different aspects with RED. It's all about how it fits your story.

There are standards, sure, and for many reasons I agree that in mainstream hollywood there should be. But I don't think anyone is going to resist seeing a picture because it's 2.35 instead of 2.39. Could you tell the difference?
 
Are you saying Stuart that you can adjust the RedCode on the camera to take aspect ratio into consideration while shooting? Therefore discarding the information outside that area and maximize your image quality/resolution needed?

If so you guys are doing a fabulous job!

I know you're saying that you are working with NLE's to make sure all this works but how about capture cards? Any plans to have the redcode working with something like blackmagic or AJA?
 
Haakon, I didn't say it's important to shoot 2.39 instead of 2.35 or the opposite. Of course, it's similar. My interest is to know which is the standard because I must shoot in one or another.

And I can't understand how it's possible someone that worked on Superman has been posted those numbers 1920 x 818 (2.35)?? When you're saying the standard was changed to 2.39 since 1970 but people still refer to it as 2.35 just due to convention?!

http://www.dvxuser.com/V6/showpost.php?p=566032&postcount=12
 
Last edited:
Gotcha. Sorry, wasn't trying to single you out - but I can see where the confusion arised (now you're getting me confused!) :) Simplest answer, the standard today is 2.39. That is for movies acquired on film, however; perhaps the Genesis has reverted back to older standards? I didn't see the original post about Superman, so I'm not sure what to make of those numbers (for some reason the link you pasted doesn't appear to be working correctly). Interestingly enough, FCP has a "matte" preset for 2.35:1 but not 2.39:1. That seems odd given the state of things... so that begs the question, is it really 2.35 or actually 2.39 with the common 2.35 label (wouldn't that just be the icing on the cake)? And if it is indeed 2.35:1, why not include a 2.39:1 in addition/instead as that has been the standard for over 30 years?

It's true that many people still say "2-3-5" because that's what they're used to and "2-3-9" just never caught on, I guess. There are lots of people who say "1-8-5" when they're talking about HD releases as well, because it's more common in the [motion picture] industry and rolls off the toungue more nicely than "16 by 9" - even though they're a bit different. You're right, there should be more clarity here. Unfortunately, the only people that really care about these descriptors are people in the industry themselves where there are all KINDS of strange names for things that don't make sense!
 
Last edited:
Haakon said:
the standard today is 2.39. That is for movies acquired on film, however; perhaps the Genesis has reverted back to older standards? I didn't see the original post about Superman, so I'm not sure what to make of those numbers (for some reason the link you pasted doesn't appear to be working correctly).
Click to expand...
Now, it's already fixed:

http://www.dvxuser.com/V6/showpost.php?p=566032&postcount=12

Interestingly enough, FCP has a "matte" preset for 2.35:1 but not 2.39:1. That seems odd given the state of things... so that begs the question, is it really 2.35 or actually 2.39 with the common 2.35 label (wouldn't that just be the icing on the cake)? And if it is indeed 2.35:1, why not include a 2.39:1 in addition/instead as that has been the standard for over 30 years?
Click to expand...
I'm afraid there is a lot of ignorance here. These guys of the computers don't understand anything about motion picture. Maybe that's the reason. I just hope that will not be the case of the digital film gear people.

It's true that many people still say "2-3-5" because that's what they're used to and "2-3-9" just never caught on, I guess.
Click to expand...
Nevertheless, I don't think so. I already read 2.35 instead 2.39 from the RED team mouth or Silicon Imaging guys where they are putting: 1920 x 818. For Stuart English, it will be 1920 x 816:

http://www.dvxuser.com/V6/showpost.php?p=545054&postcount=3

So this answers any doubt. And if it's not the case of apprehensible labeling, it's just ridiculous.
 
Last edited:
Zig_Zigman said:
So does "windowing" 2k takes up less hard drive space than "sampled" 2k assuming 24fps?
Click to expand...
Shouldn't have any affect on the hard drive space. If the numbers of pixels are the same and the compression scheme used is the same, then it won't take up any more or less space. The difference between windowing and downrezzing (sampling) just has an affect on the picture itself (do you want a center crop that can only acheive 16mm DOF and isn't sampled, or do you want an image that's been downscaled from the full s35 4K frame) and the data rate the camera can handle (in windowed mode you can crank a full 120fps off the chip whereas in the sampled mode you can "only" go up to 60). Space shouldn't be affected, though, all other things being equal.
 
Gordon Prince said:
Nevertheless, I don't think so. I already read 2.35 instead 2.39 from the RED team mouth or Silicon Imaging guys where they are putting: 1920 x 818.
Click to expand...
I didn't mean in terms of RED, I just meant in the industry. The majority of films made in the widescreen aspect are 2.39 now, but people still commonly refer to it as 2.35. As for RED choosing to actually use 2.35, that's fine with me... I will most likely shoot using the full 16x9 and crop to taste.
 
I wouldn't say RED is choosing 2.35 rather 2.39. As I posted it seems more an error from the usual labeling on the subject. That's why I think it will be important to call their attention to this. Sooner than later.
 
You must log in or register to reply here.
Back
Top