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 :)

Why...

Question

Question

Do we need a 30mm X 15mm sensor in the camera to achieve a true 4K in the final product? Whether it’s to Film or whatever source afterwards?

Humberto Rivera
 
Define "true 4k" - you can get that now with the RED One, albeit at very, very low MTF.

Graeme
 
But isn’t the current R1 camera prints out bellow 4K, and does the 5K camera with the 30mm X 15mm sensor prints out in the final film output print out to 4K? That’s basically my question. That’s whether it illusionary or whatever.

Humberto Rivera
 
The higher the resolution of the detail, the lower the MTF, necessarily, to avoid nasty aliasing. So the MTF of the RED One drops to practically zero as it approaches 4k. We see good detail out to around 3.2k. So, on a 5k camera, MTF will drop to practically zero around 5k, but you'll still get good detail out to 4k.

If you engineer your camera the other way, to get good MTF at your rated resolution, you'll either be having nasty aliasing or nasty electronic edge sharpness, or both, and they both really contribute to the "video look".

Graeme
 
Jannard said:
Just to put this all in context, I have posted before that the highest measured resolution of a 4K film scan from S35 film is 3.1/3.2K (one of the major studios tech department told us that last year). And that was from ISO 50 film.

Jim
Click to expand...

I was just trying to explain this to someone last night. It's amazing how many experienced film guys, (DP's, Editors etc.) don't realize that 4K capture is higher resolution than 35mm film.

Thanks to you Jim, I think they are learning....
 
Graeme, thank-you! As I understand it, a 5K camera will give me a great 4K resolution on the final film print! That’s pretty much the resolution needed today for film release. So the S35-X would be the best option today.

Humberto Rivera
 
Trevor Meier said:
Graeme,

The general consensus has been A/D sampling at the mean of the voltage variability of a photosite. Any thought gone into "oversampling" in the vertical/voltage domain? E.G. 16, 20 or even 24 bits at the A/D level? In theory wouldn't this give an improvement to the quality of the noise, and a reduction of 'noise aliasing' at the noise boundary?

Especially in a linear/RAW system this could have an interesting impact on the noise characteristics of the system.
Click to expand...

My understanding is that because of the nature of a binary system, that you only need an extra bit over the maximum possible sensitivity of a sample, in order to avoid aliasing and clear the nyquist limit, so adding 4 or 5 bits is just going to give you much slower processing speed and much larger files for no benefit at all(and even potentially introducing more noise because of slight inconsistencies in voltage levels during transmission). Sampling nothing is still sampling nothing even if you call it oversampling.
 
Steve Sanacore said:
I was just trying to explain this to someone last night. It's amazing how many experienced film guys, (DP's, Editors etc.) don't realize that 4K capture is higher resolution than 35mm film.
Click to expand...

I guess that's why so many people in-the-know are so blown away seeing Red 4K material projected in 4K....a near religious experience by all accounts!!! :Angel_anim:

[not to mention no weave, scratches, sparkle or even those curious cues to change reels :thumbsup:]
 
The bit depth you quantize at is separate from the sample rate - they're two different things.

Graeme

Kenneth Elkington said:
My understanding is that because of the nature of a binary system, that you only need an extra bit over the maximum possible sensitivity of a sample, in order to avoid aliasing and clear the nyquist limit, so adding 4 or 5 bits is just going to give you much slower processing speed and much larger files for no benefit at all(and even potentially introducing more noise because of slight inconsistencies in voltage levels during transmission). Sampling nothing is still sampling nothing even if you call it oversampling.
Click to expand...
 
Steve Sanacore said:
I was just trying to explain this to someone last night. It's amazing how many experienced film guys, (DP's, Editors etc.) don't realize that 4K capture is higher resolution than 35mm film.
Click to expand...

In terms of delivery, a film (s35) with projection delivers at best 850 lines in a typical theatre (vertical). On paper, 1000 ish lines at best. Often its much lower that that.

www.etconsult.com/papers/Technical Issues in Cinema Resolution.pdf
www.cst.fr/IMG/pdf/35mm_resolution_english.pdf

Much has been written about this.

Digital acquisition and projection can be far superior in terms of maintaining resolution.
Of course there are other advantages to shooting film. Hopefully that gap will close in the next couple of years.
 
Steve Sanacore said:
I was just trying to explain this to someone last night. It's amazing how many experienced film guys, (DP's, Editors etc.) don't realize that 4K capture is higher resolution than 35mm film...
Click to expand...

David Wyatt said:
I guess that's why so many people in-the-know are so blown away seeing Red 4K material projected in 4K....a near religious experience by all accounts!!! ...
Click to expand...

But as per previous posts in this thread: Red One 4K resolution = 50 ASA 35mm film neg. = (approx.) 3.2K (?)
 
Trevor Meier said:
Any thought gone into "oversampling" in the vertical/voltage domain? E.G. 16, 20 or even 24 bits at the A/D level?
Click to expand...

The majority of still cameras, and even some cinema cameras (e.g. DALSA) already oversample much more than is useful in the A/D.

Trevor Meier said:
In theory wouldn't this give an improvement to the quality of the noise, and a reduction of 'noise aliasing' at the noise boundary?
Click to expand...

No. For an explanation, may I suggest:

 
JBB said:
But as per previous posts in this thread: Red One 4K resolution = 50 ASA 35mm film neg. = (approx.) 3.2K (?)
Click to expand...

How much of a typical feature is shot on 50 ASA film ? If you are comparing film projection to digital projection, I believe there is significant resolution lost in the process of making and projecting the final film print. As I recall the typical release print only achieves a true resolution equivalent to 720p video, not even 1080p, although various other factors (grain, gate weave, etc) make the comparison very apples-and-oranges.
 
The concept of a 1-bit gigapixel photon counter is interesting - similar to how all audio ADCs start life as a 1-bit difference engine that's then quantized for PCM audio. For now though it's still a research project.

What I'm talking about is not more bits to get more DR. I'm wondering if greater quantization can improve the practical sensitivity of the camera by improving the capture of the noise. Right now the camera has 11.3 stops of DR but for most the bottom two stops aren't usable in practice because of the noise characteristics. Improving the precision of the quantization near the bottom of the sensors sensitivity could possibly be of benefit.
 
Jannard said:
Just to put this all in context, I have posted before that the highest measured resolution of a 4K film scan from S35 film is 3.1/3.2K (one of the major studios tech department told us that last year). And that was from ISO 50 film.

Jim
Click to expand...

Actually Quantel also had this figured out in September. Not sure the relevance, but i remember the conversation with and them saying
"actually, the 4K is 3.2K debayered"
 
Daniel Browning said:
The majority of still cameras, and even some cinema cameras (e.g. DALSA) already oversample much more than is useful in the A/D.



No. For an explanation, may I suggest:

Click to expand...

Thanks for those references - very educational.

This thread started with Jim's explanation of the 30 x 15 mm sensor size, which has led to much discussion of 32 mm image circles. The current dilemma for Red owners is whether to use our opportunity to buy RPP lenses. My experience suggests that what I need is not more resolution, but more dynamic range. So ideally I would want the Monstro chip with it's greater DR. So the crucial question is: (hard to answer for a camera that does not yet exist, but driven by the immanent purchasing decision) will the RPP cover 5K of the Monstro sensor? With it's larger size and pixel size, what is the image circle required for 5K on Monstro?
 
Irrelevant maybe...
but I still wonder when the epic will be released?
is 3 months a wishful thinking?
 
You must log in or register to reply here.
Back
Top