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 :)

RED B16 vs F23/Panavision

C

cinepost35

Banned
Joined
Dec 23, 2007
Messages
169
Reaction score
0
Points
0
I know it's not out yet but how do you think Build 16 ultimatley will compare differentley now with an F23 or Panavision all cameras output to 2k space for print?

Should build 16 push Red beyond any reasonable doubt or argument?

It's clear that it is a huge step forward, Sony / Panavision obviously have never seemed to offer this kind of software upgrade for there cameras with this kind of improvement .
 
  • Thread starter
  • Banned
  • #3
OR....
F23 / Panavision scaled through a Lowry/DTS process to 4k and Red to 4K
projected in IMAX .

U23D as an example was shot with Sony 950's to Sony SRW1's decks and scaled to 4k and looked beautiful in IMAX 3D.

How do you think RED in B16 over 15 would look different or better if U23D would have been shot with Reds and output today vs the Sony technology that was available at the time U23D was made?

It feels like the true test is in a projected image and not on a HD monitor.
 
There have been software upgrades to F23's and Panavision Genesis cameras, and the Arri D20 especially -- you just don't hear about them generally. And the Genesis is a rental-only camera, so any improvements are done internally at Panavision, not announced publically every time it happens.

Also, to some extent, those cameras were released closer to their final "ultimate build level", whatever you want to call it, they didn't beta-test them with owners to the level that RED has with their cameras, so there is probably less potential for improvement through software updates. Also, with separate sensors for RGB with the F23, or a simple RGB striping system with the Genesis, there isn't a debayering process needed with the potential for improvements in how that it is done. The RED and the Arri-D20, on the other hand, are continually improving the quality of their debayering process to reduce noise and improve dynamic range. When you have separate sensors for RGB, ways of reducing system noise tend to come from other physical methods, like improving the circuitry, etc.

The Sony F900 went through various software and firmware upgrades ever since it came out in 2000, three major ones including a new DSP, preamp. Then it was upgraded completely as the F900R.
 
skipdivils said:
how about build 16 vs. Sony F35......

But the REAL cage match will be the EPIC taking on all comers....
Click to expand...

this will probably be a fairer fight since these other cameras have been
field tested and improved over years of use and R&D.

With what is essetially Red2 in Epic we will see, I feel, the giant killer in Digi
cinema at just the second incarnation.
 
  • Thread starter
  • Banned
  • #7
Is it fair to say that both RED B16 and Arri D20 latest software would offer overall advantages in picture quality over the Sony / Panavision family of products if one was considering all possibilities for a film going to 4K budget aside.

In my previous post if you could use any camera available without concern for budget what would hold up the best going to 15/70 IMAX or 4k Digital for projection. Considering all went through some Lowry/DTS type process to get each to the best 4K space possible coming from each cameras best ouput stream for the lab and then projecting.

One example I put out is the U23D IMAX as one example if that film was shot today in Red B16 how would that compare theoretically to the Sony F950 recording to Sony SR format and scaled to 4K for output.

Just like the statement 4K makes better 2K it seems that if it holds up on an IMAX screen it will look great on a traditional size screen.
 
David Mullen ASC said:
The Sony F900 went through various software and firmware upgrades ever since it came out in 2000, three major ones including a new DSP, preamp. Then it was upgraded completely as the F900R.
Click to expand...

True, and one of those upgrades (the /3) was, in many circles, considered as radical an improvement as Red's Build 16 is being touted.
 
mmost said:
True, and one of those upgrades (the /3) was, in many circles, considered as radical an improvement as Red's Build 16 is being touted.
Click to expand...

Yes, but that was both a software and a hardware upgrade in the case of the F900/3.

For IMAX release, I think resolution is an important factor and in theory, the RED, whatever Build version you have, should have the edge over the HD resolution of the Sony and Panavision cameras, though you'd want to test that. But in theory, the RED is a little over 3K in practical resolution compared to 1.9K of an HD camera. In practice, there are other factors affecting apparent sharpness and how it looks enlarged, so test.
 
I agree in testing to really see what your getting. But for discusion purposes it seems RED B16 with a much cleaner image should look much better than HD scaled to 4K and using something like the Lowry/DTS process would have more image data to work with than HD and in theory yield a better image.
 
I don't fully understand the math, but a software genius just told me that when the Genesis was being developed, testing was done to see if striping or Bayer pattern was better. The result suggested that a Bayer sensor had to be 5K in resolution to compare to the striped sensor, which has 2K red, 2K green, and 2K blue. Hence, the Epic, which should pass that threshold.
 
I do fully understand the math, and bayer at 4k gives, much greater luma resolution and at least equal chroma resolution as the striped approach. There's a reason no DSLR uses the striped approach, because it's an inferior colour filter array pattern. Sure, you get equal numbers in RGB, but you get a LOWER number!!!
 
It seems no matter what, an RGB striped sensor recording to 4:4:4 1080P HD can only be 1.9K resolution per channel max, no matter how good or how many pixels the sensor has. To me, it seems that the main advantage of a striped sensor is not really the picture quality over a Bayer filtered sensor's image, just the simplicity and speed of decoding/converting it into RGB. Essentially all you need is a "dumb" algorithm, not a complex one as Bayer cameras use. So part of the quality of a Bayer image is tied to the quality of the debayering process being used, and it's a process that can be improved over time.

Remember that a RED camera only has to do a partial "quickie" debayering at 4:2:2 720P quality for monitoring, whereas the Genesis has to convert the signal from the RGB striped filter to its highest possible quality in real time for recording to 4:4:4 HD. This was probably part of the reason for avoiding a Bayer pattern, that and designing the system around recording to HDCAM-SR.

I will say, however, as John Galt will probably remind me whenever I run into him next, that pixel resolution is not the same thing as sharpness. And as Graeme will also remind me, besides sharpness you have to factor in aliasing artifacts. So it's a complicated subject as to what constitutes a "better" image for large screen projection, and what artifacts one finds more objectionable than others. One might look casually at a F23 or Genesis image and feel it looks similar in "sharpness" to a RED image, but once you start looking at each channel individually, look for aliasing artifacts, look at MTF, look at edge characteristics, you may prefer aspects of one image over another for different reasons.
 
David Mullen ASC said:
It seems no matter what, ...
Click to expand...

David, your technical prowess never ceases to amaze me. Sometimes it appears that you are not an English major. I have one quick comment. Much has been said about the real-time deBayer of raw data on this forum. I can vouch that hardware based solutions are, though complicated, but possible that do very good deBayer in real-time. For one, we built one and it gives very good results for the intended purpose.
 
I'm sure it's possible, and I'm sure that RED will keep improving the live debayering in their cameras. Afterall, the Arri D20 is normally recorded to HDCAM-SR, so it has to do a high-quality live debayering from its 2.8K RAW Bayer sensor at 4:4:4 1080P quality.
 
Mind you, the D20 is considerably heavier and larger than a RED ONE, and I'm sure that beefier DSP hardware makes up a lot of that.
 
David Mullen ASC said:
I will say, however, as John Galt will probably remind me whenever I run into him next, that pixel resolution is not the same thing as sharpness.
Click to expand...

Likewise, having sharpness equal pixel resolution is not necessarily ideal also. Having more pixels than sharpness lets you get a beautiful and well defined edge quality rather than a harshly sharp edge.

First Light Films said:
I don't fully understand the math, but a software genius just told me that when the Genesis was being developed, testing was done to see if striping or Bayer pattern was better.
Click to expand...

Software geniuses can make mistakes too. If he was using an inadequate debayer, he could easily come to the wrong conclusion.
 
It's Lowry not Lowery

It's Lowry not Lowery

Say whatever you like about us, just spell the name correctly. ;)
 
First Light Films said:
I don't fully understand the math, but a software genius just told me that when the Genesis was being developed, testing was done to see if striping or Bayer pattern was better.
Click to expand...

Being a software genius myself :shiftyph34r:, I would say that absence of evidence is not evidence of absence.

It is inconclusive without knowing what testing has been done. If I flip a coin 5 times and get "heads" every time , does it prove that it is a "heads only" coin? If I flip it 500 times and get "heads" every time, now I'm pretty convinced that it is "heads only" coin.
 
You must log in or register to reply here.
Back
Top