What is 6K?

[Humberto Rivera]

Phil you’re information is most useful to understand the full range of 6K Digital Cinema as always! However I wish you might include the 6000 X 3000 measurements on your different visual representations since it shrink the Image Circle by a “Full Point”, and that makes them useable with the Master Primes, or Leica Summilux C, or Angenieux Glass at the wider focal length. Thank, you’re work is always the best!

Humberto Rivera

 

[Andy Roberts]

I am not an expert on printing, but you certainly don't need to print to 300dpi all the time. A good source file can produce a large print, even at lower dpi.

Many large format vendors ask for images to only be 150ppi.

 

[Mark Andersen]

Many large format vendors ask for images to only be 150ppi.

Its about viewing distance, catalogs and magazines are 300PPI, sometimes higher. Billboards can be as low as 15PPI. I could see posters for public display going in the 150 range. 300 is minimum for fine photographic prints to be seen up close, often 400PPI is preferred. However, a 19M files will scale up nicely in photoshop. The point is the stills of the dragon are pretty much useful for any purpose.

 

[Justin Marx]

Boy u crayze.. Love these facts!!

 

[Martin Stevens]

Thanks Phil.

 

[Shadi Abu Sharar]

Thanks Phil

 

[Kennan Ward]

Phil,

Great contribution, and education here!

Thank you!

Kennan & Karen Ward

 

[Dr. Sassi]

Thank you Phil for your ongoing supply of support, leaves only one question ;o) ... may we have your 6K graphic on T-Shirts? )

Have a great day

 

[Tehben Dean]

Great info.

Now we just need a chart to show how 6k affects processor power in post in relation to 5k, and 4k...
5k had a significant hit on performance over 4k, I can only imagine what 6k will do. But the unfortunately since being able to shoot 5k, I never shoot 4k... and I bet it will be the same once moving to dragon. :\

 

[Nick Fury]

Hi Phil!

As a corollary to your perspicuity, as regards 6K's imaging advantages, what is your analysis and plan, going forward, for an editing and archive system, please? Are one of these http://www.reduser.net/forum/showth...roup-ArGest-Asset-(-Edit)-and-Archive-Bundles! , or similar, on your short list? Don't want to derail your thread, so if you are inclined to respond, wherever is kewl.

Thank you very much

Fury

 

[Phil Holland]

Heh. Thanks everybody. I'm not that crazy

Many large format vendors ask for images to only be 150ppi.

Yep. Good point and added this:
At 150 DPI you can make a 40.96x21.067 inch print from a full 6K Dragon image.

Its about viewing distance, catalogs and magazines are 300PPI, sometimes higher. Billboards can be as low as 15PPI. I could see posters for public display going in the 150 range. 300 is minimum for fine photographic prints to be seen up close, often 400PPI is preferred. However, a 19M files will scale up nicely in photoshop. The point is the stills of the dragon are pretty much useful for any purpose.

Yep. I've done billboard and banner work in the past. Can't remember the print size at E3, but it was probably the largest photo I had printed. It was on the side of the LA Convention Center and the optimal viewing distance was easily 75 feet. I got to see it on the ground and it wasn't really pixels, but globs due to a up scaling algorithm used.

Thank you Phil for your ongoing supply of support, leaves only one question ;o) ... may we have your 6K graphic on T-Shirts? )

Heh. You can use that one for whatever you'd like.

As a corollary to your perspicuity, as regards 6K's imaging advantages, what is your analysis and plan, going forward, for an editing and archive system, please?

My general plan doesn't really change. I have hard drives (triple redundancy) and LTO that are pretty standard fair. Studios and clients that I work for nearly all use LTO as well.

I am keeping a keen eye out for any new higher capacity burnable disk solution. I can see a day of setting up a dual or triple array to archive that way. Just not yet.

Great contribution, and education here!

Thanks Kennan. Take it easy on those little fellas with that big honkin lens Love the chaos in those shots.

 

[Alain M]

Hey Phil! if you'd make video tuts, I am sure you would be loved by people with tons of hits! heheheheh. Very easy to understand for amateurs, intermediate, pros, masters, etc...


Awesome explanation!!!

Thanks!

Alain M

 

[Will Keir]

Thanks Phil, I appreciate the work and clarity you put into these diagrams. Keep up the good work.

 

[Nick Pasquariello]

Something's been bugging me for a bit, and I thought I knew the answer, but rather than assume, I'd like to be sure:

In the case of Red (and other Bayer pattern cameras), the pixel count is physical pixels on the sensor, correct? Meaning a grid of 4 Green Red Blue Green photovoltaic cells = 4 pixels. This data is then debayered (i.e. put through an algorithm in order to combine data from neighboring pixels intelligently thus creating a, in this case, 6K viewable image). Since each photovoltaic cell only works with one of the 3 primary colors, yet still counts as a "pixel" once the neighboring data has been integrated, then you sort of end up with less resolution than 6K, in a sense; each pixel isn't measured, it's partially measured and partially calculated. So, in a sense, 6K footage would be somewhat downscaled during the debayer, and then upscaled in order to display the 6K image in 6K. This alone would explain why the best 4K comes from a 5K sensor or larger: when debayered and combined you end up with the ideal down-sampled image.

Is this correct? Or are there more than 6K photovoltaic cells, and the "6K" nomenclature = image post-debayer and not photovoltaic cells. If this is correct, then a Scarlet shooting at 4K 24fps would look best at 3.2k or less, which I think I read somewhere on the forum a while back anyway.

 

[Mike Nagel]

Something's been bugging me for a bit, and I thought I knew the answer, but rather than assume, I'd like to be sure:

In the case of Red (and other Bayer pattern cameras), the pixel count is physical pixels on the sensor, correct? Meaning a grid of 4 Green Red Blue Green photovoltaic cells = 4 pixels. This data is then debayered (i.e. put through an algorithm in order to combine data from neighboring pixels intelligently thus creating a, in this case, 6K viewable image). Since each photovoltaic cell only works with one of the 3 primary colors, yet still counts as a "pixel" once the neighboring data has been integrated, then you sort of end up with less resolution than 6K, in a sense; each pixel isn't measured, it's partially measured and partially calculated. So, in a sense, 6K footage would be somewhat downscaled during the debayer, and then upscaled in order to display the 6K image in 6K. This alone would explain why the best 4K comes from a 5K sensor or larger: when debayered and combined you end up with the ideal down-sampled image.

Is this correct? Or are there more than 6K photovoltaic cells, and the "6K" nomenclature = image post-debayer and not photovoltaic cells. If this is correct, then a Scarlet shooting at 4K 24fps would look best at 3.2k or less, which I think I read somewhere on the forum a while back anyway.

I had similar thoughts / questions / misconceptions and Graeme answered that here:

http://www.reduser.net/forum/showthread.php?103919-Debayer-Export-w-o-Scaling

- M

 

[Nick Pasquariello]

So if I'm reading that correctly, 6K is the count of the bayer patterned photovoltaic cells?

 

[Mike Nagel]

So if I'm reading that correctly, 6K is the count of the bayer patterned photovoltaic cells?

6K is the physical pixel count, which obviously is utilized via a Bayer pattern on the CMOS sensor.

80% is roughly the measured resolution, depending on other factors as well, such as lens and color science...

maybe the new Dragon color science improves things even further...

 

[Phil Holland]

This alone would explain why the best 4K comes from a 5K sensor or larger: when debayered and combined you end up with the ideal down-sampled image.

Mike's got the right idea. Essentially correct if you're focus is a mathematically typical debayer down sample. You want 80% of the physical pixels debayered down to your target resolution. Which is why the Mysterium-X sensor was designed specifically with the resolution of 5120x2700, which comes out to 4096x2160 for a full debayer.

Dragon furthers this a bit with a couple things. It's a 66.7% down sample/debayer to 4K.

However, REDCODE RAW is interesting stuff and depending on what you do with the material you can indeed create stunning 6K images with good sharpening practices. I've done this with MX at 5K, you'll be able to do it with Dragon at 6K. One other thing to keep in mind is Dragon's smaller pixel size, which should help pull out a bit more spatial detail. And you'll have control over really how "crisp" you want that final image.

 

[Gunleik Groven]

Thanks Phil. Appreciate it!

G

 

[Mike Nagel]

Mike's got the right idea. Essentially correct if you're focus is a mathematically typical debayer down sample. You want 80% of the physical pixels debayered down to your target resolution. Which is why the Mysterium-X sensor was designed specifically with the resolution of 5120x2700, which comes out to 4096x2160 for a full debayer.

That's a statement that I don't fully get (yet)... probably b/c I'm envisioning this process step by step instead of an all-in-one process, which might yield different results...

STEP BY STEP PROCESS:

debayer always comes before scaling, IF you decide to scale... decompression > debayer > scaling (optional)

so when you debayer 5K MX footage, you get 5K footage (an image with physical pixel count of 5K) with roughly 80% measured resolution, which would be roughly 4K of measured resolution WITHIN that 5K image.

So let's say we debayered Epic footage to 5K (in whatever codec) and now decide to (after debayer) scale that debayered 5K image to 4K. Just b/c the image has roughly 80% measured resolution that doesn't mean you will get 100% measured resolution when scaling to 80% of the pixel resolution. A down scaling algorithm surely chooses the "best pixels" so measured resolution will improve, but you'll never end up at 100% measured resolution.

Or do you ? ;-)

Besides other factor's like lens etc and that the 80% is just a rough number, than it would also come down to which software (RCX | PP | AE | Resolve | etc) does the down scaling (different sw, different results), and which down scaling parameter / options are being used...

so from this perspective, I can't really see the 4K out of 5K argument, but I can totally see that a downscaled image will always improve results...


ALL IN ONE PROCESS:

Maybe you get better results, if you setup to debayer and scale at the same time, maybe the SDK does a "better" / smarter debayer as it knows now it will downscale and can disregard "inferior" pixels right away, but somebody would have to confirm that...

If that is the case, here's a follow up question:

does it yield better results to start off with debayering / scaling (all-in-one) to 80% pixel resolution to get the most superior debayering results and then (optionally) upscale later ?

Upscaling afterwards could be done with any sw, and you might get better results using a 3rd party sw...

So, for the example of the Scarlet @ 4K:

captured 4K > debayer / scale to 3.2K (to get "best" debayer results) > then (optionally) upscale with whatever sw you like to 4K


What am I missing ? ;-)

- M