4k live... ( 4k Still from Red One is up )

Well, I'll chime in with my 2c congrats as well -
I'm really looking forward to being challenged by this new medium. It's going to be really hard to get away with ANYTHING shooting on RED.
The really exciting thing will be when REDfest 1 comes along (and you know it will...) Fifteen movies in the desert with nineteen year olds? Yeah, I guess I could live with that. Hell, I'm in Cali, I DO live with that.
Oh, what's the point - a picture's worth a thousand of these and my keyboard would give out.
 
Deadzone said:
Is there any reason a low-budget guy like myself would have to go 4K?
Click to expand...

Film-outs and green screen are a couple I can think of. But if you don't need these, 2k or even 1080p will be more than enough.

I know just because of the relatively reasonable data rate, I am going to go ahead with 4k even for projects that are destined for dvd release.
 
baquajim said:
Film-outs and green screen are a couple I can think of. .
Click to expand...

IMHO, dynamic range and low compression are more important than resolution for green screen work (given that most compositors soften their keys a bit anyway). But I think Jim and Graeme are all over that one. If Redcode delivers the stated compression specs WITHOUT breaking the image up into blocky chunks with artifacts (like jpeg for example) then we're laughing. I have wondered if there would/could be specific Redcode options for optimizing green screen / blue screen shooting - perhaps allocating more data for green/blue channel??? Not sure.
 
I think to optimize Redcode for greenscreen, you would devote equal quality/bandwidth to the red, green, and blue channels.

Intuitively, it makes more sense to allocate more to the green channel... however, I don't think the math works out that way. i.e. think of a grayscale... it has green values from 0-255.
What you are really keying off of is the relationship between R:G:B... this ratio contains the hue and saturation. Actual greenscreen would have high G:R and G:B ratios... this is what you're keying off of essentially. *There are different algorithms, but I believe they all key off the r:g:b ratio.

2- Wavelet-compressed RAW encoding should offer significant advantages of DCT / Y'CbCr encoding.
DCT versus wavelet: Wavelet encoding tends to remove noise (and image details). DCT tends to introduce noise (i.e. mosquito noise on high-contrast edges)... which gives you those boiling keys.
Wavelet encoding might give you ringing on edges, which may not key well. But this depends on the functions you use... it might be good to choose a function that won't produce ringing on edges. *Ok, I haven't tested this.

RAW versus Y'CbCr: With RAW encoding, you have lower quantization noise in the dark portions of an image (and usually dark areas are very hard to key, although noise is a big factor). Y'CbCr has lots of rounding error and is wasteful since most of the color space represents illegal colors the camera can't produce. RAW = no color space waste.

With Y'CbCr encoding, typically encoders optimize greater bandwidth for luma instead of chroma (this is correct for non-greenscreen use)... whereas for greenscreen use, you want equal bandwidth for luma and chroma (an ideal keyer would key off both luma and chroma; any of the "3-D" keyers work off luma+chroma).

But basically, RAW recording and wavelet compression should offer significant advantages for keying.

3- This is somewhat speculative, but I would think that the best way to optimize for greenscreen would be in carefully choosing the functions for all the resampling/sampling-related functions. Ringing on edges would be bad.

Some resampling functions do introduce ringing (phase shift errors) intentionally. This is because they measure aliasing a certain way... and the only function that reduces anti-aliasing will have phase shift errors. This has to do with the uncertainty principle- in a (sampled) signal, you can't be certain of both frequency + position (phase).
In practice, it might be better to live with some aliasing and have no ringing.

Perhaps a visual explanation would make more sense:
Look at
http://en.wikipedia.org/wiki/Anti-aliasing
In the checkerboard pictures, the (c) image shows a sinc filter. The sinc filter has 'perfect' anti-aliasing properties, but if you look closely there is ringing / phase shift errors on the checkboard (zoom in). The top edge is also erroneous.

*That example may not be the best, since there are other effects the images didn't account for (i.e. gamma correction / the image getting darker in the middle of c).

In Red, you have sampling-related functions in:
A- The sensor. The sensor has to implement optical low pass filtering. i.e. blurry glass. If there is too much blurinnes, you want to conteract it (un-blur it digitally). Optical low pass filtering is necessary to reduce aliasing of various sorts.
B- Going from ~5K --> 4K. (I don't think this will be applicable.)
C- In wavelet compression.
 
Last edited:
Thanks Glen. I'm going to have to sober up, and then re-read your post. So much to learn....
P.S. - just noticed your another Toronto guy. Go Leafs!
 
boothba said:
Weird - I just noticed that Red.com briefly posted the two shot of the girls smoking as a 4K file, but now it's gone (back to the bubble girl). I didn't manage to downlaod in time. Was that a tease?
Click to expand...
yeah yeah c'mon jim, post these and other goodies!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
 
Emanuel said:
&I think both of you have more information here:

LINK
Click to expand...

I agree more with Jim's 3.2K guesstimation. The 6K mantra comes from Kodak reps and the like, but I've worked on too much IMAX stuff to believe that.
 
boothba said:
I agree more with Jim's 3.2K guesstimation.
Click to expand...
This guesstimation isn't coming from Jim. I think it was people from Panavision that gave this info to him, I believe. But you can ask to Jim.
 
Emanuel said:
This guessestimation isn't coming from Jim. I think it was people from Panavision that gave this info to him, I believe. But you can ask to Jim.
Click to expand...

I thought it was coming from the (refreshingly honest) folks at the studio (Warner Bros?) following his first 4K presentation. I should have qualified that statement.
 
Jannard said:
We did a remedial dead pixel correction (not precision) as stated on the website. It didn't get them all. This week lots of things get addressed, including DPC. You are also looking at an 8-bit jpeg, which does NOT look as good as the 16-bit tiff.

Jim
Click to expand...


Thats pretty much what i figured. I cant believe how sensitive to these thigns ive become. I used to not care.
 
boothba said:
The only question now is how much you'll have to degrade the Red image (add grain, film weave, etc.) to make it look identical to 35mm film. Based on these early tests, they should WELCOME a shootout.
Click to expand...
Film is wonderful, but I don't believe it's the "be all and end all." Digital has already handedly exceeded the use of traditional film in the still world; there is no question in my mind that it will ultimately do the same in the realm of motion picture. Just takes a bit longer as technology plays catch-up.

In any case, you won't see me "degrading" my RED footage. As much as people want to say that grain "adds character" to a shot, in most cases, it's just because grain is an inherent part of celluloid film and it's what we've always seen on the screen. In 50 years when a whole new generation of kids who are used to seeing nothing but beautiful, clean images begin to make movies, they're going to look back on films of today and wonder why everything we shot looks so "dirty."
 
boothba said:
I thought it was coming from the (refreshingly honest) folks at the studio (Warner Bros?) following his first 4K presentation. I should have qualified that statement.
Click to expand...
I stand corrected you're probably right.
 
Haakon said:
In any case, you won't see me "degrading" my RED footage. As much as people want to say that grain "adds character" to a shot, in most cases, it's just because grain is an inherent part of celluloid film and it's what we've always seen on the screen.
Click to expand...

I'm with you... I can't think of a better imaging future than what's coming with that butter smooth "24fps DSLR" look the RED team promises. I want clean, clean, clean, for technical AND esthetic reasons. However, there will be some "educational adjustment" for the viewers, as this story shows...

... Years and years ago, I was the first guy in my region to use a Spirit DataCine for 35mm to tape transfer. Everyone else was using the regional Ranks with their noisy blue channels and nasty gate weave. I'd finally convinced a local production company to piggy-back some "B" roll 35mm from one of their jobs onto my transfer just so they could check out the sweet Spirit lovin'.

They get the footage back, and thread it up and start to watch... I'm with them, and it was CLEAN and SOLID and the 90fps tennis stuff was absolutely gorgeous! Or so I thought...

... A couple weeks pass and I see their finished job. They had went back and "Ranked" it. It was now nasty and grainy and bouncy. What the... ??? Come to find out they had shown the Spirit footage to the client, who HATED it. Said it looked too much like VIDEO, heck, in fact, thought that it was video.

Regards,
 
People forget how clean modern 35mm film is. The technology is so good on 35. This clean red footage really does look great.


And BTW Mr. JAnnard, i apologize for bringing up the dead pixel thing, i missed the note on the page on your site. i got so excited i just clicked the image without reading, lol. Sorry to restate it.
 
Wow it is funny I'm on my 17" Imac at home and all I can see is a lust the chin some lips and thats it 4K is really really big. I think I'm going to hafe to order two 30" to work with this. I think it is a good reasone to get them. I have always wanted one just need a reason.
 
You must log in or register to reply here.
Back
Top