Why do some Sample Videos appear Washed out?

[lynx]

Hi,

I was just looking at the sample footage under the "Gallery" tab on the Red website, and was a tad surprised that the Peter Jackson footage as well as the one just below it (that looks like two Milkmaids), appear a bit low in contrast and color saturation.

Does anyone know whether this was intentional or whether Red Output in "Non-Studio" conditions appears just like this?

Thank in advance for your answers.

Regards,
Lynx

[Cail Young]

My understanding is that it was the grading choice by Jackson to make his film look like it does; as for Milkgirls who knows what Jim Jannard prefers (about contrast, I mean...).

Given that the camera is shooting 12-bit, low contrast in an output image isn't an issue because you can just stretch it in a colour correction or grading tool.

[Graeme Nattress]

Generally, the higher the dynamic range, the lower in contrast the image will look. Also, monitor calibration will play a part too.

Graeme

[Nick Shaw]

Don't forget it will look different depending on your monitor setup. Are you looking at the images with your monitor set to 2.2 gamma and in a Colorsync aware application that honours the embedded profile?

[lynx]

I'm using an Apple Macbook Pro .. and I haven't changed the default Color Settings/ profile.

As a followup to Graeme's statement, then we shouldn't just be considering Dynamic Range by itself as a parameter of quality... Contrast + Faithfulness of Color Reproduction must also factor into the picture. Right?

If so, could someone post some information by way of a comparison in terms of "Image Quality" between the Red and say 35mm Kodak Vision 2 Super 35 Film Stock.

Thanks in advance again..

Regards,
lynx

[Brook Willard]

First, change your ColorSync profile to match the Adobe 1998 / 2.2 gamma viewing conditions that the RED site recommends.

With regards to contrast and dynamic range, I imagine that the RED team wished to put as much of the viewable dynamic range into the JPG images as possible. That is, since they wanted the images to show all of the detail captured, they may have created lower contrast images . This is just speculation, of course. A 12-bit image flattened to 8-bit in a linear fashion will always look washed out.

Had they released a contrasty image, people would've complained about lack of dynamic range. Since the images on their site were made to show the technical abilities of the camera instead of the artistic elements of a story... the images may look "flat" when compared to a final look that you may have chosen instead.

Case in point: If you took a Kodak 52__ stock and scanned it to a DPX file with all of the negative's available dynamic range, the image would look milky and flat. It is the color correction/grading/timing you do afterwards that makes it look "right" - whether your definition of "right" is contrasty, moody, flat, whatever.

If I had to make something up, I'd say the closest equivalent would be 5299. It's Kodak's DI stock... not meant for printing. If you've ever seen a straight print of the stock [I did it for fun once...], it's the ugliest, milkiest thing you've ever seen. But when you grade it later... it looks gorgeous. This is, of course, a blind comparison from a stock I'm familiar with to a camera I've never personally shot with. Don't take this point seriously.

That's the beauty with this camera. You capture [i]everything the camera is capable of capturing; you're never locked into a look. Only afterwards do you get fancy with the spices and make the image you really want... with all of your original data still available. It's like DI on steroids. :)

[lynx]

Very interesting. We did a DI for a Feature I shot recently and I do remember that the Scanned (LOG) Images looked milky. The Lab used a Look Up Table (LUT) to convert them into a Linear Space.

Could someone please confirm : that the images that Red has posted are LOG output down-converted to 8bit Motion Jpeg?

I have heard that with Final Cut Studio 2 some (or much of this) may be possible. Could someone please recommend a good book / white papers that would allow me to get my feet wet with this... what is a LOG format/ what is a Linear format/ sample LUTs/ and the LOG to Linear Conversion process....?

Many Thanks in advance.

Regards,
lynx

[Brook Willard]

Could someone please confirm : that the images that Red has posted are LOG output down-converted to 8bit Motion Jpeg?


Just for the record, I don't recall saying this.

[Graeme Nattress]

The images have never been near log. We also need to get some terminology straight. You say linear, but you don't mean it. You're meaning implies a video gamma, which if you've ever looked at the graph, is no-where near linear.

When we say linear, as in, our sensor is linear or we record the linear output of the sensor, we mean totally straight line linear. Linear to light.

Log is easy, there's a logarithmic curve.

So, we have 3 things,

Linear, as in linear light,

video gamma, which is confusingly called "linear" by many, due to it being perceptually (to an extent) linear,

and log, which is log.

Graeme

[lynx]

Dear God, I really need a book/ white paper on this... I had no idea that Linear referred to / was used in two ways.

But to get back to what I was asking earlier, how good is Red in Capturing Color with Fidelity? i.e even though the image may look milky, is all the inherent color captured accurately (as compared to Kodak Vision 2 for example).

Many Thanks,
Lynx

[Mike Devlin]

Steve Shaw's DI Guide does a nice job of explaining the relationship between gamma and apparent contrast. He compares Video (Viper and F900) versus Film (OCN and release print). You will notice in his sample images that film original camera negative is the most "washed out" since it is the image with the highest dynamic range. The release print has less dynamic range and appears high contrast.

Steve is a fan of log gamma curves, so Graeme would probably disagree with a few of Steve's positions, but the basics are the same.

Steve's website is http://www.digitalpraxis.net/ and then go to the section on technical papers. The direct link to download his DI Guide is
http://www.digitalpraxis.net/zippdf/di-guide.pdf

[Stephen Gentle]

On my computer, the videos look completely washed out in Quicktime player, but really good in VideoLan...

Is this because Quicktime is set up wrong? I can't find any setting to change this anywhere.

[Joel Kaye]

how good is Red in Capturing Color with Fidelity? i.e even though the image may look milky,

Better than anything else close to it's price range. Search for milk girls and you'll find all kinds of different color corrections applied to it.

Straight off the sensor the image is raw and it's going to look flat... but a LOT of data is there to be manipulated and that's the most important thing.

As far as accuracy goes - my guess is it'll be more accurate than anything else in it's price range... but perhaps we'll have to wait for those tests.

[dalemccready]

Quicktime player doesn't colour manage so it'll look washed out more on your monitor if it is set to 1.8 gamma. Watch it with After Effects or FCP and it should colour manage to some degree. Or change your monitor setting to 2.2 as Brook has already mentioned.

[Gunleik Groven]

The easy thing to do, is to download one of the 4k images and play with them in photoshop.

Plenty vibrant in a sec or 2 if you ask me.
Or de-saturated
or...


edit:
And I thought this place was dead... 5 simultanious answers -;)

whatever -;)
Gunleik

[Álex Montoya]

Yep, the problem is most probably of the Quicktime player.

[Bruce Allen]

Check your video overlay settings for one thing if you're using a PC. I have mine set up to match my Avid settings and it unfortunately means that all 90% of movie trailers, etc look milky too...

Or your problem could be totally different - like all of those happy times when Quicktime would use different gamma settings whether you rendered your After Effects project on a Mac or a PC ;) Or the gamma shifts in Final Cut 1...

I was rendering something on a PC today and the black levels were hugely different depending on whether I rendered with Photo-JPEG or PNG compression. I think Apple deliberately cripples it, the same way they deliberately make HD video playback slower in the Windows version.

And they wonder why we still like stacks of frames! When someone makes a file format which:
a) doesn't screw up simple things like rendering consistently gamma across platforms
and
b) can easily allow multiple computers to write to it at the same time with low chance of corruption (eg for multi-machine rendering)
...then we'll change over!

Of course, on the other hand we could just convince Apple to make their Finder able to display folders with thousands of frames in them without crashing... you'd think Jobs would have figured out this was important by now, what with the whole Pixar thing...

Bringing it more back on topic, I sincerely hope that RedCode footage doesn't suffer from all of these PC / Mac Quicktime gamma inconsistency horrors...

Bruce Allen
www.boacinema.com

[dalemccready]

feel better now Bruce? ;)

...but I'm SO with you on the finder being a 'cup of cold sick' when it comes to dealing with large numbers of files like DPXs and such. Go out grab a coffee while the finder tries to even look at the files...better yet, grab lunch, come back, force quit the finder and use unix instead...

ahhh, now you've got me doing it! ;)

[Graeme Nattress]

Log is fine as long as you completely linearise it before doing any work on it. That's make linear, not make video gamma :-)

Only issue there is if you have super-whites, which you get with how most set the white point with log encoding, you apply a gamma curve, which a lot of people use for colour correction, and guess what, super-whites get darker as mid-tones get brighter......

Graeme