Investigation into Digital Cinema Cameras

[Ashley]

Currently, I’m in the middle of my dissertation, investigating the imaging quality of digital cinema cameras. The reason for this post is to enquire as to whether anyone knows where I can obtain a copy of the Red One and Epic technical data i.e. Colour and Resolution chart tests.

The information from the tests would enable the development of my dissertation to take a more relevant direction that is applicable with current technology. Any help would be greatly appreciated.

 

[paulherrin]

Hey Ashley, I would probably start by trying a search for "test charts" on the forum here. There may be some other people that remember where things are better. Also, they'll probably ask for your full name asking for things like that. Best of luck!

 

[Dan Hudgins]

No fixed value

No fixed value

I can obtain a copy of the Red One and Epic technical data i.e. Colour and Resolution chart tests.

There is no fixed value for the "resolution" of a digital cinema camera since you can adjust the MTF contrast over a wide range with the OLPF compensation and other sharpen tools.

Also the lens used has a big impact on the apperent resolution and S/N ratio, since the more sharpen you use the more noise you see.

You can adjust the sharpen and lens to make a good camera give low readings on a test chart, and you can fiddle with the de-Bayer adjustments and make a poor camera read much higher on test chart values.

The de-Bayer used in RED cameras is quite good, but its base adjustment is set a bit soft to help prevent motion aliasing. Looking at a single still frame of a test chart is a poor way to judge a Digital Cinema camera, you need to look at moving images played back without extra compression on a full resolution monitor or projector. Without the motion you cannot judge the aliasing vs. sharpness vs. MTF at various frequencies, etc.

To avoid aliasing in digital images you need at least one half pixel on each side of a single pixel detail, and the diagonal of pixels is 1.4x the HV dimension so you should divide the H pixels by two to get the visual resolution at higher MTF values (not the >1% MTF value). Also Bayer filters only have 25% fill on RED and BLUE pixels, so to get 2 pixels (0.5+1+0.5) with in red and blue, you need to divide the H pixels by 4. That gives you a useful number to compare to a film image not processed through a DI or film recorder (which use digital images).

So a 4K image gives about 1K solid "resolution" without much aliasing.

1920 gives about 480 solid "resolution" without much aliasing.

A 35mm film-to-film print reads between about 1280 and 960 under good conditions at resonable MTF as projected off a 4th generation print.

You can goose the HF of digital images or the film scans used in DI, but the aliasing will go up.

In other words, the 4-5K RED images are about equal to 35mm film, Bayer 1920x1080 HD cameras need to be oversharpened to look like they are about the same (and so can show aliasing), and 3 chip HD cameras can just about need the 35mm level because they have 100% pixel fill in all three colors, Bayer is only 50% in the Green and 25% in the Red and Blue, which is why Bayer cameras need an OLPF filter to blur details over several pixels before processing to reduce chroma Moire.

3 chip cameras have issues with the lens and the prism block, and cannot use "normal" lenses, only ones made for the prism block, which they may or may not be fully corrected for.

ARRI used 2880x1620 and so does Scarlet to get 1920x1080 so the oversample gives better results than with no oversample then reduction by 1/1.5x.

The image area used in film recorders can be 1828x1556, 1280x1.4=1792 so 1828 is not enough to not show "any" aliasing (1280x2=2560), one reason that some films are finished with a "4K" DI, and all digital images will show aliasing if they are oversharpened. Bigger and less sharp is better than smaller and more sharp.

If you look at Kodak's D-LAD image it looks like it was sharpened as there are sharpen artifacts along the frame-lines and image, but moving images, at least un-compressed ones, look sharper when moving than viewed as still images, with film its about 2x as sharp when projected at speed than when viewed still frame.

As for the color, it does not matter much what a test chart taken by the camera looks like, since most projects re-grade the frames in good software that makes the end result look different that the output of REDCINE-X or the RED SDK anyway. With a RAW camera there is no fixed way anything the camera outputs will look. The look changes with the software and firmware builds as well. Even the noise level changes with de-noise in post and temporial noise reduction.

 

[Ashley]

Hi,

Thanks for the reply. I’m not use to posting threads or for that matter using forums. I don’t know how to change or add personal details on this site. If any one knows let me know. But here is a little information about me:

My name is Ashley James Grant. I’m studying a BSc in Film Production Technology at Stafford University (uk).

 

[Ashley]

Dan Hudgins,

Thanks for the reply.

Ashley

 

[David Birdy]

Changing information

Changing information

Ashley,

Welcome to the Reduser forum. You found a great place for information as this forum is full of knowledgeable professionals

You can edit your details under the "quicklink" tab in the upper right corner of the page.

Dave

 

[Graeme Nattress]

Dan - way too much dodgy info in that post. ALL cameras need OLPFs regardless of sensor arrangement for instance.

Graeme

 

[Ashley Grant]

Hi,
I was the one who originally posted the message, and i thank all of you for the response. I was banned for reasons unbeknown to me and have just re-registered. I'm not use to forums so is there anything i need to provide so that i don't get banned again.

David Birdy, I'm sorry, i could not access the message you sent. Could you send it to me again.

I'm sorry for any inconvenience.

Ashley

 

[Dylan Macleod CSC]

Geoff Boyle posts various tests from different digital cameras at his site
cinematography.net
It is a great forum that will have useful info for you.

 

[David Rasberry]

Hi,
I was the one who originally posted the message, and i thank all of you for the response. I was banned for reasons unbeknown to me and have just re-registered. I'm not use to forums so is there anything i need to provide so that i don't get banned again.

Ashley

Strictly the real names issue. Nothing you said or did. Welcome to Reduser. :thumbup1:

 

[Alexander Ibrahim]

Dan - way too much dodgy info in that post. ALL cameras need OLPFs regardless of sensor arrangement for instance.

Graeme

I certainly don't understand this stuff as well as you Graeme, but it seems to me that Dan's point about OLPF was that all cameras have them, and it makes it a bit more difficult to get at the truth about resolution, and then he went on to explain why.

Perhaps you are objecting to the part where he talks about why Bayer sensors use OLPF ... but that merely does not address other sensor arrangements, it doesn't invalidate his earlier statement.

The truth is that you're real input on this topic would be a real asset, if you are willing to do so instead of just taking potshots at others posts.

Even without saying a word about RED you could add a lot. Just thinking and writing here about other digital camera systems would be an immense addition. Heck you could probably just write a list of public links and reference materials and help us out a ton.

This is a point of discussion where a cinematographer might gain little or no benefit, but I happen to have a technical education. I was training to be Spock you see. (Only half joking. My majors were computer science, electrical engineering, math and physics.) I'm terribly out of date and out of practice- but I'm still curious.