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 :)

4K or 5K process

E

Eric J Kosinski

Well-known member
Joined
Jul 26, 2011
Messages
372
Reaction score
0
Points
0
I'm having a bit of a hard time finding out how exactly the resolution of RED DIGITAL is so good, using a super 35mm sensor? My main question is;

Is it only/mainly software/firmware pulling this off? I've heard the RED ONE has about 8million pixels, and the EPIC about 14. Sony f3 has approx 12 megapixel chip only capable of 2K or less, and an Arri Alexa around the same from what I hear at Abel Cine tech; DSLR, etc, etc has the same or larger megapixel image sensors. Why is it possible for RED to accomplish higher resolution vs these other cameras with "better" sensor pixel ratios?

Is this done through technology in firmware? Is it interpolation, etc? If so why hasn't any other manufacturer offered this as a firmware upgrade? Any help will be appreciated.
 
The RED process, I believe, can be outlined fairly simply.

1) Start with a high quality, high resolution imager. Epic-X uses a 14 (13.9) MP imager with very low latency, high dynamic range, and many options in terms of crops and framerates.
2) Write that to media with REDCODE. Don't scale it, don't sharpen it, don't use massively visually lossy compression, and don't lock the image using a traditional "video" codec.
3) Use great (free) software like REDCINE-X PRO and/or great color and gamma like REDcolor2 and REDlogFilm to begin to interpret the image.
4) Give the user control of what they want to do creatively - sharpening, denoising, grain, grading, etc.

That's how I see it; I'm sure Graeme, Jarred, or Jim could put it much better.


And Bob, I believe that Alan Roberts (and several others) soundly disagree with the 3.36MP assessment on the F3.
 
Alan makes the following assessment:

Sensitivity was not measured directly. The specification claims it to be T/11 at ISO800, and since ISO800 corresponds to 0dB gain, this means that the sensitivity is very similar to that of a 3-sensor 2⁄3” sensor, which in turn implies that the pixels are about 5μm square. Given that the sensor is ‘super 35mm’ size, it must be 24x13.5mm. For the pixels to be 5μm spaced, the sensor width must be about 4800 pixels, making the sensor approximately 4,800x2,700. This fits reasonably well with the estimations in section 1.2.1, and means that the sensor has approximately 12.9 Megapixels, typical of a digital stills camera. It also explains why there is little or no coloured aliasing, and why the red, green and blue signals all have the same resolution and aliasing.

His assumption is that the pixels are 5µm spaced....only an assumption and is somewhat disputed my Sony's PDF.

Alastair Chapman, as I recall lists the sensor in the 3.4 range...I will look for his assessment.

Thanks,

Bob
 
Very interesting stuff, Bob. I've never read much about the F3, but did take a gander at Alan's white paper. Looks like he may have guessed incorrectly!
 
Look up oversampling...

Look up oversampling...

IMO the proof is in the actual measured resolution of the image any camera creates. For starters, a zone plate or other reliable chart can be shot and the resulting image analyzed. Due to things like Bayer pattern filter schemes, optical low pass filters, processing electronics, compression ratio and type, etc the theoretical resolution of any device is dependent on more than just counting photosites.

Now if you care to play some chess ;-), try measuring the resolution of objects moving horizontally across the frame at various speeds, then vertically, then - for extra credit - high contrast subjects moving diagonally across the frame. Then repeat at various shutter angles (speeds). Having fun yet?

There are some very clever ways to deliver high resolution images from electro-optical systems of widely varying types. With the EPIC, RED has managed to engineer an amazing combination of form factor, file size, recording media, compression and sensor technology able to deliver high resolution records of real world scenes that include fast moving subjects. Nothing that is even remotely portable can match it (at least nothing that's not classified ;-).

FWIW the F3 is the most impressive camera to come out of Sony in years. The F23, F35 and the 900 series CineAltas were were all overpriced pigs.

Now the F65... we should know more very soon...

Cheers - #19
 
Bob Moore said:
Numbers are a bit high.

Sony F3 at 3.36 approximate...large photosite allow great sensitivity....

See this thread:

http://www.dvxuser.com/V6/showthrea...on.-Enough-for-free-aliasing-1000TV-lines-444

http://www.dvxuser.com/V6/showthread.php?237005-Sony-official-brochure-for-the-PMW-F3-(just-released)


Bob
Click to expand...

I have seen a video or article directly from Sony advertising the benefits of the Sony F3 against DSLR (specifically the 5D) in which a Sony Rep stated they use a 12+ megapixel sensor, which by having less pixels offer larger photosites, but never heard it was as low as 3.36 MP ??

Either way, I still don't understand how even a 5D Mark II with 22 Megapixels can not pull 5K + if the sensor of a Red (super 35) can ?? Is it the software that is doing this ?? If so does anyone know why others haven't offered this as an option ? Or is it simply "HYPE" interpolating pixels and the other manufacturers are just not using that as a sales tool ?
 
Blair S. Paulsen said:
IMO the proof is in the actual measured resolution of the image any camera creates. For starters, a zone plate or other reliable chart can be shot and the resulting image analyzed. Due to things like Bayer pattern filter schemes, optical low pass filters, processing electronics, compression ratio and type, etc the theoretical resolution of any device is dependent on more than just counting photosites.

Now if you care to play some chess ;-), try measuring the resolution of objects moving horizontally across the frame at various speeds, then vertically, then - for extra credit - high contrast subjects moving diagonally across the frame. Then repeat at various shutter angles (speeds). Having fun yet?

There are some very clever ways to deliver high resolution images from electro-optical systems of widely varying types. With the EPIC, RED has managed to engineer an amazing combination of form factor, file size, recording media, compression and sensor technology able to deliver high resolution records of real world scenes that include fast moving subjects. Nothing that is even remotely portable can match it (at least nothing that's not classified ;-).

FWIW the F3 is the most impressive camera to come out of Sony in years. The F23, F35 and the 900 series CineAltas were were all overpriced pigs.

Now the F65... we should know more very soon...

Cheers - #19
Click to expand...

That definitely gets me closer. So basically it is more than the sensor; it is more like the whole package but specifically the technology from RED I guess.
 
Eric,

3.4 K.

Major size photo sensors with a great ASIC for conversion...beautiful HD files.

I came within a hairs breadth of buying a F3...if it had a better codec I would have one today. I realized that
anything from Sony less than the PMW-500 was crippled for financial reasons...always make the customer ante-up
for the next thing. Poor business model...bad ethics.

That is what sets RED apart...

Bob
 
Bob Moore said:
Eric,

3.4 K.

Major size photo sensors with a great ASIC for conversion...beautiful HD files.

I came within a hairs breadth of buying a F3...if it had a better codec I would have one today. I realized that
anything from Sony less than the PMW-500 was crippled for financial reasons...always make the customer ante-up
for the next thing. Poor business model...bad ethics.

That is what sets RED apart...

Bob
Click to expand...

So the Sony F 3 is actually 3.4 K ? Interesting. Is this the reason why there are claims that RED "fakes" the real resolution for hype? Or should I ask WHY don't all the camera companys use the K rating to show off the sensor resolution for higher sales like RED?

So would Sony F3 be close to the footage of the RED ONE / EPIC when projected on the big screen or does the codec mess quality up to loose a lot of detail/resolution? I have to admit the F3 is attractive, especially for the quality for the price, but if RED EPIC S arrives soon, that will be the choice for sure.
 
F3 is a 1080p camera. Don't confuse pixelcount of a sensor with the actual output or recording format. Even with the S-Log upgrade and Cinedeck, still 1080p output.

...unless I've got my info crossed somewhere.
 
Jarek Zabczynski said:
F3 is a 1080p camera. Don't confuse pixelcount of a sensor with the actual output or recording format. Even with the S-Log upgrade and Cinedeck, still 1080p output.

...unless I've got my info crossed somewhere.
Click to expand...

OK, so correct me if I'm wrong, the Sony f3 has a 3.4 K sensor but resolves to 2K (or less 1920 x 1080) but the RED sensor is 4K (red one) or 5K (epic) and resolves to true resolution of the Native sensor ?? Please help as this is where I'm confused?
 
Every large single sensor camera I know of uses a sensor with a resolution in excess of it's captured format resolution - except cameras that shoot RAW or compressed RAW files. This is because of the sensor design and the use of different pixel patterns - Bayer, striped, etc. These are not 100% effective at transmitting the real world data into pixels because each photosite can only read as red, green , or blue. So oversampling is used to compensate; have lots of pixels and your image becomes more directionally correct.

The early muck about RED revolved around the concepts of sensor resolution, captured format resolution, and effective resolution. What people were saying is that even though the camera had a 4K sensor and captured a 4K file, it wasn't really a 4K camera because it couldn't actually resolve 4K after debayering. This is, of course, inherent. The Alexa can capture a 2.8 K image, but it certainly won't resolve 2.8K after debayering. And DSLRs can capture 5+K images - but they don't resolve their sensor resolution either. Again, this comes back to the fact that the pixels on the sensor have to be interpreted to create the final image, resulting in a built in and expected loss of resolving ability. Hence the EPIC, which shoots 5K so that it can resolve a great 4K image.

Notice also, that I mentioned the captured file resolution - this is important. Say the F3 does have a 3.4K sensor (which, from that reading, it very likely does) - why don't you get a 3.4K file out of the camera? Because Sony has chosen to have the image processed inside the camera and output in a much smaller and lighter format, capable of being written out to much more cheaply available media. This is a compromise that results in 1) the use of a video codec and 2) the loss of data from the original image. A 3.4k sensor could sensibly resolve close to 2.5K and make a really amazing 1080p. But if that data is thrown away for the sake of ease of use and in-camera processing, its gone forever. A similarity can be observed with the Alexa. Normally, the camera uses it's 2.8K sensor to produce a 1080p file which is written out to the SxS card. But if you attach an expensive recording module and enable the recording of raw data, you can get that 2.8K image.

So what "K" is any given camera? It's sensor size? It's recording format? Or it's resolved resolution?

Also, I get the feeling that Dan will probably come along soon and make a post so informative I'll have to spend three days digesting it. Can, open. Worms, everywhere.
 
Eric J said:
OK, so correct me if I'm wrong, the Sony f3 has a 3.4 K sensor but resolves to 2K (or less 1920 x 1080) but the RED sensor is 4K (red one) or 5K (epic) and resolves to true resolution of the Native sensor ?? Please help as this is where I'm confused?
Click to expand...

You're wondering why a 5D MkII can't capture 24fps RAW video at the full pixel count of the Canon full frame chip?
Processing power. The 5D just doesn't have the processor horsepower to make it happen.

Overall, though, don't confuse pixel count with resolution. '5K' refers to the fact that the sensor has 5,000+ pixels available horizontally. "Resolution" refers to how much detail can be, ahem, "resolved" by a given imaging system and that number depends on a huge number of factors (like Bayer vs. Non-Bayer, OLPF characteristics, the lens, etc.

Resolution is not equal to pixel count.
 
Michael Olsen said:
Every large single sensor camera I know of uses a sensor with a resolution in excess of it's captured format resolution - except cameras that shoot RAW or compressed RAW files. This is because of the sensor design and the use of different pixel patterns - Bayer, striped, etc. These are not 100% effective at transmitting the real world data into pixels because each photosite can only read as red, green , or blue. So oversampling is used to compensate; have lots of pixels and your image becomes more directionally correct.

The early muck about RED revolved around the concepts of sensor resolution, captured format resolution, and effective resolution. What people were saying is that even though the camera had a 4K sensor and captured a 4K file, it wasn't really a 4K camera because it couldn't actually resolve 4K after debayering. This is, of course, inherent. The Alexa can capture a 2.8 K image, but it certainly won't resolve 2.8K after debayering. And DSLRs can capture 5+K images - but they don't resolve their sensor resolution either. Again, this comes back to the fact that the pixels on the sensor have to be interpreted to create the final image, resulting in a built in and expected loss of resolving ability. Hence the EPIC, which shoots 5K so that it can resolve a great 4K image.

Notice also, that I mentioned the captured file resolution - this is important. Say the F3 does have a 3.4K sensor (which, from that reading, it very likely does) - why don't you get a 3.4K file out of the camera? Because Sony has chosen to have the image processed inside the camera and output in a much smaller and lighter format, capable of being written out to much more cheaply available media. This is a compromise that results in 1) the use of a video codec and 2) the loss of data from the original image. A 3.4k sensor could sensibly resolve close to 2.5K and make a really amazing 1080p. But if that data is thrown away for the sake of ease of use and in-camera processing, its gone forever. A similarity can be observed with the Alexa. Normally, the camera uses it's 2.8K sensor to produce a 1080p file which is written out to the SxS card. But if you attach an expensive recording module and enable the recording of raw data, you can get that 2.8K image.

So what "K" is any given camera? It's sensor size? It's recording format? Or it's resolved resolution?

Also, I get the feeling that Dan will probably come along soon and make a post so informative I'll have to spend three days digesting it. Can, open. Worms, everywhere.
Click to expand...

This was extremely helpful in my quest. This board is freakin' amazing. Post a question and within an hour understand the entire concept. Love it. Thanks guys. I was under this assumption, but to have the knowledge helped a lot. Thanks again for the detailed info.
 
Eric J said:
Either way, I still don't understand how even a 5D Mark II with 22 Megapixels can not pull 5K + if the sensor of a Red (super 35) can ?? Is it the software that is doing this ?? If so does anyone know why others haven't offered this as an option ? Or is it simply "HYPE" interpolating pixels and the other manufacturers are just not using that as a sales tool ?
Click to expand...

I think in the case of the 5D it does take larger then 5k still images with it's 22MP sensor but it's processor is not capable of capturing more then 3-4 of them per second. So even with new firmware it is incapable of recording 4k video. Why the F3 doesn't I do no know but probably has to do with the cameras processor, media, and the codec that sony came up with (or didn't come up with) among other things.
 
For what it's worth, the F3 is a fantastic 1080p camera. Recently used it on a couple commercials and it held up great. That said, I'd never buy one over an Epic. ;)
 
You must log in or register to reply here.
Back
Top