PDA

View Full Version : Color chart test shoot?



Jan Wenstad
12-09-2013, 12:27 AM
What would be extremely helpful in understanding what the Dragon sensor's technical advancements/limitations would be the following:

- Color chart shot in a controlled environment with both cameras set to 2k
- 5 sec record on iso 400 stepping up to 2000
- Adjust the f-stop as you move up the iso scale, but keep them the same on both cameras.

If you are in LA, I would be happy to come help out. It shouldn't take more than a couple of hours to do.

Thanks

Gunleik Groven
12-09-2013, 12:36 AM
This would be a rather meaningless test on a few accounts.

NOT saying colorcharts are meaningless, as they are not.
But this testprocedure would not give a lot of valid info...

Jan Wenstad
12-09-2013, 12:07 PM
I have to completely disagree with you on this one. I will list one that would be very helpful:

- grain patterns can be analyzed per color at the pixel level without down res'ing the footage to eliminate it

Your tests are great at illustrating worst case scenarios, but I am almost positive that all of us don't intend on shooting our projects by the hip.

It would be great to see what the Dragon can do and what to look out for when my camera get's updated.

Also, it would be interesting to hear how this test wouldn't be helpful to do? For me, knowing what happens to "green" when shooting a green screen shoot is pretty important.

Gunleik Groven
12-09-2013, 12:57 PM
Here is why.

Practically, adjusting the iso does not alter the exposure.
Recording 2k will make you neglect the extra oversampling in the Dragon which contributes to lower the noisefloor.

It is "evening the field", but it does not gove much practical user info for most scenarios I see people shoot at.

One thing which could be a good thing, if I understand your suggestion right, would be to shoot a greenscreen under different light temperatures, and tungsten with/without gels. Has been done a few of those tests on Epic, but would be interesting to see on Dragon and in a comparison, especially taking into account the dragons lower IR response.

But again... If you want to get any practical info on keyability, it would probably be good to take resolution into account, too...

If you do not take the different oversmpling ratios into account, you are basically shooting the dragon in the chin on one of its strengths.

Color renderition being the other, should not suffer. but it would look more noisy than it in practical terms is.

That is why I say that this procedure will not give a lot of "valid data"...

Jan Wenstad
12-09-2013, 01:28 PM
I do have to agree with you that one of the strengths of Dragon is being able to shoot in 6k, in turn making "grain" into a "who cares." Unless for some insane reason you were hired by the State Government to shoot in the middle of the night in the Desert and they wanted to see everything in the frame, which happened... the footage spent the next week being de-noised. This odd job doesn't happen often, but the blue channel has definitely caused me some problems.

Sorry to get off point...

The purpose of the test is to stress test the camera for grain... We all know that the other benefits are undeniable, but understanding the grain and knowing my limitations as a DP is most important when choosing the right tools for the job. If I know that the Dragon's sensor begins to introduce more grain at 1000 iso that causes more time to be spent in post to make the footage pristine, then so be it... I would like to know what to expect without the other benefits that the camera has clouding how I use the tool.

I think this is a very important test in so many ways... Greenscreens / and the occasional strange night shoot is where I can deliver expectations with true confidence.

The point of the test is to take an unbiased approach to test the technical limitations when considering "grain." The adding of other elements (i.e. 6k) is just clouding the water and we can't see what is really happening.

Gunleik Groven
12-09-2013, 01:33 PM
You know what.

I am sorry for my remarks.
Please test in any way you want and share your results.

is little you know about what kind of projects I am involved with from what I post here, as little do I know if this test suits your purpose.

There were two key elements in the procedure you described, that I am a bit allergical to, as they have created a lot of confusion for a lot of productions I have had to support.

I guess that is partly why I try to push the tools visibly in ways many do in practical terms, but think they can hide "in post" or wherever, and ends up in a mess.

As long as you have a purpose and know what you look for, any test can be valid.

Best of luck

G

Jan Wenstad
12-09-2013, 03:59 PM
My intention was not to remark poorly on your test, they were very helpful. But it would be great to go under the hood to really see where the new sensor sensitivity stands without software intervention. I need to know where the starting line is before the race starts. That's where all the videos promoting the camera fell short... all of them seemed to showcase the person behind the camera and not the camera by it's own accord.

Anyone else in LA who has access to a Dragon? I'm available to run some tests.

paulherrin
12-09-2013, 04:28 PM
i think it's a good idea, but i think it's silly to rush it before the first huge color science update drops

go ahead and start planning, that way you'll be ready to go right when it does

Blair S. Paulsen
12-09-2013, 08:02 PM
I agree with Paul, judging Dragon color rendition with MX color science is a waste of time.

Cheers - #19

Brian Merlen
12-10-2013, 03:07 PM
well its a valid test for now, but yes everything will change once the dragon color and gamma is out. but the same can be said of all tests happening now, so i figure any tests are fine, even if its on pre dragon color science. i think the important part is just making sure people realize that the color science isn't done and were betaing it up sort of... i am sure the dragon LUTs will be excellent in all due time...

Eric Haase
12-13-2013, 11:29 PM
I agree with Paul, judging Dragon color rendition with MX color science is a waste of time.

Cheers - #19

Why? That's what's currently available. People are shooting with it. Sometimes it's good to know where you've been to understand where you're going...

Marc Wielage
12-14-2013, 01:25 AM
I agree with Paul, judging Dragon color rendition with MX color science is a waste of time.
I dunno about a waste, but I am curious to see where Red goes with the color science with Dragon. The differences between RedColor2 and RedColor3 weren't blockbuster changes, but it was an improvement.

Gunleik Groven
12-14-2013, 01:30 AM
Now... Putting on the speculative glasses. Warning! ;)

I think I see what might happen especially with DragonGamma, and it has the potential to be pretty rad compared to RG3. RedlogFilm will probably not change much, though, and performs incredibly with Dragon.
DragonColor. Now that makes me curious.
Colors are already very good. But I see how they can change somewhat.
Graeme has a tendency to pull wildcards, so I will just wait and see.
I have a hunch what might be in there, too, but what I am thinking of is pretty rad compared to what I have seen in REDs and others cameras.
Then again.
Dragon is pretty rad.

As to the test described, I am not sure what use it has.
But testing now, as for short form quick turnaround productions "now is now". I can see the use of that. Obviously....

Jan Wenstad
12-16-2013, 11:58 AM
Look... The point of the test is not to judge Red's color science or how it takes the RAW camera information and translates it.

The test should be the same settings on both cameras. I was attempting to set up a test to eliminate anything that could effect the image in either direction.

Let's just compare the sensors MX and Dragon. How sensitive are these new chips? What kinda noise do they produce with different colors? How much noise do they produce? And how do all of these relate to what we have been shooting on.

It is a very easy test to do and will be very informative to those who care about such a thing in their production. Lets do a chip to chip comparison without software intervention with the upcoming updates.

I'm still available to do the test in Los Angeles, anytime... Just need a Dragon.

Filip Orlandic
12-16-2013, 12:04 PM
How sensitive are these new chips?

According to mr. Bill Bennett, ASC Dragon sensor is ISO 200 native. That's what he was told by RED.

Blair S. Paulsen
12-16-2013, 12:46 PM
I guess we'll have to agree to disagree Jan/Eric. I respect the commitment to learning and greater understanding in any case.

There are a number of factors, many interactive, that form the basis of the color matrices involved. IMO, one of the keys is the dyes used in the CFA. Precise alignment between the characteristics of the dyes and the de-bayering algorithm are the foundation of color linearity. Until we have the color science optimized for Dragon it's my belief that all you can learn is how to most convincingly untwist the mismatch. YMMV.

Cheers - #19