Blackmagic Forum

Hi there,
at the moment i am using a cheap color chart from amazon:
https://www.amazon.co.uk/gp/product/B00 ... UTF8&psc=1

But after a while it is not working properly anymore. Or maybe it has never worked properly??

-Anybody has had bad experience with cheap color cards?

-And which color card would you recommend me to buy?

Thanx in front
Greetz Lenny

Hi,
I do not want to be derogatory but why buy a color cards that has not been certified by BMD if you intend to use it with BMD Davinci Resolve?

That said, unless it's been improved, I've never been able to get the color card matching to work with any kind of satisfactory result even when using an X-Rite card.

PeterMoretti wrote:I've never been able to get the color card matching to work with any kind of satisfactory result

It depends on what you expect as a result. It does not automatically compute a 3DLUT-like correction that places all the color patches on the calibrated pattern within 0% of the known target values. It does attempt to neutralize the grey scale, depending on how well lit the chart itself is, and then displays an error value, showing where the spectral deviations are. If you added another node to your correction tree, improved the secondary hue values, or in fact, the grey scale deviations, then added another node and re-performed the chart correction to generate another chart/reference test, then recursively fixed your intermediate node, you could, in theory, generate a Look-Up that would compensate for all the lighting inconsistency, lens flare, non-uniform spectral density of your lighting source(s), the transfer function of the light-gathering sensor, its PROM/error compensation, the camera's internal matrixing, the rounding errors in the media codec, couple of hundred other factors...

For that setup.

Then you would get to do that all over again if you change the lighting, exposure, lens, metadata...

Charts are great for a couple of things. People who have never done this before can benefit, and those who have the experience and skill to do it right and understand what it is that the cards signify can communicate many intentions with judicious use of the card and clear notes to anyone who needs to interpret the card.

They are not that useful for precision auto-correction unless the cinematographer is bullet-proof. Usually by that point, though, they will have sorted out that its just easier to bypass the tool and read the scope.

jPo, CSI

Hi there and thanx for the reply's, JP, Peter and Jean,

I do not want to be derogatory but why buy a color cards that has not been certified by BMD if you intend to use it with BMD Davinci Resolve?

I can only agree with your point. I guess as a beginner you think you can do everything a bit cheaper

That said, unless it's been improved, I've never been able to get the color card matching to work with any kind of satisfactory result even when using an X-Rite card.

Okay that is interesting to know.

Charts are great for a couple of things. People who have never done this before can benefit, and those who have the experience and skill to do it right and understand what it is that the cards signify can communicate many intentions with judicious use of the card and clear notes to anyone who needs to interpret the card.

They are not that useful for precision auto-correction unless the cinematographer is bullet-proof. Usually by that point, though, they will have sorted out that its just easier to bypass the tool and read the scope.

So my question is: for my situation would i be better of, with not working with a color chart?

I am trying to make a children's series on youtube. Merging a 3D cartoon world with a keyed out character --> filmed in my own DIY green screen studio. I am doing everything myself, so i am the cinematographer, the colorist, the compositor and the editor.



My situation: I film always the same character, with the same camera and roughly the same lighting setup (it will change some times, cause of changing from a Wide shot to a medium for example). My goal to use the color chart was, to be able to get the same Blue and Orange colors on my keyed out character every time.

But if i understand your reply properly JP, then i would be better of bypassing the tool and learn to read the scopes?

Thanx in front
Greetz Lenny

Not only that.
If you always use the same camera at a fixed color temperature and the same lighting, you shouldn't need much correction at all.

In my humble experience (and I do mean humble, lol) color charts are nice to have but once you are good enough to use the color chart, you are good enough to not need it. That said, every now and then, you do find yourself going back to the color chart and thinking "Oh wow, that was kind of helpful."

So shoot one if you have the time to and it's not a pain. No harm. As for different the workflow, I think it's a bit not there. But experiment with it and learn from other people, and you might find something that works for you.

He There,
After all the advice in this post i decided to buy the x-rite color card video.

But it still doesn't seem to work? Or i just don't understand the whole thing

Here is a shot without the color card match:


Here a shot after the color card match has been applied:


I don't understand it,....it looks completely wrong

Maybe somebody can have a look at my shot? (i exported a single .exr and put it on Google drive)

https://drive.google.com/file/d/0B8MudY ... sp=sharing

The funny thing is that i used to have a yellow outfit before and i filmed in the camera,..producing .mp4 files with 100mbps (so very compressed) And these problems didn't accure.

Now i record with the Atomos Ninja Flame --> and i have 442 8 bit prores footage with 650 mbps --> and all of a sudden these problems accure. I must be doing something wrong.

I hope somebody can help me out?
thanx in front
Greetz Lenny

Um, not ALL the advice

PeterMoretti wrote:That said, unless it's been improved, I've never been able to get the color card matching to work with any kind of satisfactory result even when using an X-Rite card.

PeterMoretti wrote:In my humble experience (and I do mean humble, lol) color charts are nice to have but once you are good enough to use the color chart, you are good enough to not need it. That said, every now and then, you do find yourself going back to the color chart and thinking "Oh wow, that was kind of helpful."

That's basically my reaction to color charts as well. The really good ones, like the DSC Charts, are actually fairly good provided they're lit correctly. For everything else, I think you can get there just manually. At worse, you could look at the manufacturer's LUT for Rec709 and see what that does for comparison.

Lennart Holterman wrote:Now i record with the Atomos Ninja Flame --> and i have 442 8 bit prores footage with 650 mbps --> and all of a sudden these problems occur. I must be doing something wrong.

Recording 8-bit is not ideal for green screen material. The composite is going to tend to "break" more when you stress is out in post.

If you can rent or borrow a better camera -- a Blackmagic Ursa Mini would be ideal -- I bet you could get a reasonable picture, even using a stock Rec709 LUT. Be aware that shooting green screen is harder than you look: you need to pull the talent away from the green background enough to avoid spill, and everything needs to be fairly evenly-lit and at a reasonable mid-level exposure.

Lennart, what camera are you shooting with?

My humble opinion:
it doesn't have to be exact and perfectly accurate. It can be what you like and what looks nice for you.

Marc Wielage wrote:the DSC Charts, are actually fairly good provided they're lit correctly.

+1 for DSC Labs charts. I carry their "one shot" chart on all shoots where I have to match wildly different cameras to each other. The advantage to the DSC one shot over other options is that the color chips are designed to match the REC709 vectors on the vector scope and yield nice big targets that are easy to read manually. It also has 4 skin tone chips along with standard greyscale. I can set a look on one camera, apply that look to the chart shot from that camera, and then use the chart shot from the other cameras to manually grade them to match.

Art Adams, who designed that chart, gives a good overview of one way to use it here:

https://www.provideocoalition.com/cameras_rough_guide_to_color_grading_with_the_new_dsc_labs_oneshot/

The comments below the article are worth reading as well.

PeterMoretti wrote:Um, not ALL the advice

PeterMoretti wrote:That said, unless it's been improved, I've never been able to get the color card matching to work with any kind of satisfactory result even when using an X-Rite card.

Yes you are right there Peter,.....i am seriously thinking of not using it at all

Dan Sherman wrote:Lennart, what camera are you shooting with?

Panasonic lumix G7 + the atomos ninja flame recorder

Lennart Holterman wrote:Panasonic lumix G7

I've got one of those so I took it down to my workshop(garage) as that has most accurate and consistent lighting (6500k CRI 85) and shot a clip.

Camera was set as follows:
[list=]
[*]standard CineLike D (no hue or contrast or saturation adjustments at all)
[*]luminance was set to 16-255
[*]White balance was set with an expo disc
[*] Internal h.264 mp4, 100mbit UHD
[*] exposure was set by using zebras so that the highlights where between 90 and 95%
[/list]

To "correct" the color I followed the manual process outlined in this video.



I'm just a hobbyist, and I'm sure some here could give the shot a lot more pop, but its looks pretty good to my eyes considering it was decent out of the camera to start with.

Ignore my grunginess, it's late and damn near 90 in the shop.

before


after


what the automated match did.

Thanx Dan,
For your testing. My footage is shot in a green screen room with tungsten lights.

Thanx for the video link,..i will follow it and see what it can do with my shot.

As a follow up, this is how much resolve says my manual correction was off by.

Pure black being off is understandable, what bothers me is the cyan and red. I assume this is a luminance issue, but every time i touch the Hue vs Lum curve the match goes to pot pretty quickly.

Code: Select all

2   4   3   2   8   3
4   3   1   2   1   2
2   2   1   2   4   3
2   3   7   2   1   0


Any of you pros have any tips to manually squeak out that last bit of "correctness"?

Forget about it, there is no 'correctness' other than what you see – on a calibrated monitor, that is.

Cameras use a complex color matrix to mimic what the human eye sees, but given the very different detectors and their sensitivity curves, there is no perfect camera, only better or worse approaches to correct colors.

Every manufacturer has a different layout of their approach, some have perfect skin tones and can't handle landscape so well, others may have problems with subtle nuances in red flowers, but get perfect blue and green.

That's why the experienced colleagues rely on their eyes, scopes and experience and use the cards only as a rough guide.

Dan Sherman wrote:I'm just a hobbyist, and I'm sure some here could give the shot a lot more pop, but its looks pretty good to my eyes considering it was decent out of the camera to start with.

A couple of power windows, secondaries, and defocus might do what you want.

Uli Plank wrote:Forget about it, there is no 'correctness' other than what you see – on a calibrated monitor, that is.

I think this is true: you can ignore the numbers and just go with how the picture looks artistically. If it looks good, it is good (assuming there's no clipping or crushing and the picture is legal for video... and the client likes it).

Marc Wielage wrote:

Uli Plank wrote:Forget about it, there is no 'correctness' other than what you see – on a calibrated monitor, that is.

I think this is true: you can ignore the numbers and just go with how the picture looks artistically. If it looks good, it is good (assuming there's no clipping or crushing and the picture is legal for video... and the client likes it).

I'm not worried so much about "correctness" for the sake of correctness. I know that will all the variables involved, just getting in the same ballpark is the goal.

I'm approaching it more from the stand point of learning how to getting footage from some cameras black box color science, to a known standard like rec 709 or 2020 or 2100 etc. I think understanding the math (i was trained as a physicist), would help me better understand how to do something like match footage between two completely different camera models and/or brands.

I get your point as a scientifically trained person.
But camera manufacturers will not tell you what their math is, that's their 'coke recipe'.
And BM will probably not tell us what their card matching math is.

So, if you are grading for a specific target, get your monitor calibrated for that target and do your aesthetic grading – that's what your audience should see in the end (if their display isn't too far off, sigh).

To match cameras, use the split frame function, the scopes and your eyes. The more you do that, the better and faster you get.