Calman 5 + C6 + Radiance Mini 3d calibration help please - Page 2

I can't see 17-Point Luminance anythere, i see 16,31,61....

16 Point Luminance with 6-Point Saturation is 286 Color Points, as you wrote this it takes about 700 Reads, with your logic a 16x16x16 Cube (4096 colors) it will take 7.372-10.240 Reads Average.

No user cares about a single click if the results are INFERIOR...

the real point here was the statement that followed right before that "to write to the LUT device twice, once to set the 3D LUT to unity and second to write the final result"... that is the improvement in convenience - NOT in quality, congrats to that... and that workflow has been around for a while...

regarding validation: all validations I ever gotten after a CM Autocal were deemed completely useless after I ran a Quick Analysis in CM immediately after the CM Autocal...

maybe you can elaborate on how we can truly validate a 5^3 LUT with CM ?

Sure it is.

17 != 255 either.

At some point the response of a display is predictable enough to create correction.

Steve Shaw himself says 5x5x5 cube that's been correctly created is enough to correct a consumer display. http://www.lightillusion.com/home_cinema_calibration.html

The fact that we can do that with fewer reads simply strengthens our argument that picking the correct points to calibrate is more advantageous than profiling large cubes and (***shock***) interpolating the results.

And with CalMAN the final Cube is already measured and validated. With Lightspace you'll have to go back and reread you points to verify the correctness of the calibration.

Yeah but they aren't

They are better.

We'll for the radiance which is only a 5^3 cube, the final set of readings is your validated, written to the box data.
If you want to brief glimpse at what' happening in between the radiance's control points our color checker run does a great job of validating.

If you want to super sample, you can always go into options and crank up the data data points.

But with a radiance, the only points either package can control are the 125 points, so validating their correctness is the most important thing.
Edited by sotti - 1/27/13 at 10:59am

Very interesting would be the validating of points which are NOT part of the 125 points of the 5^3 cube and which are not read and "calibrated / interpolated" via the software.

5x5x5 = 125, we read and validate all of them for the radiance. It's only when we go to higher density cubes that we start using additional interpolation.

they are not, I've done 1:1 comparisons between Lightspace and Calman on the EXACT SAME SETUP using a Mini.

regarding your app "validating" the cube before it applies the LUT, that is only so good, b/c as you (should) know the Mini introduces additional distortion once the LUT has been uploaded and is active... therefore one needs to read and verify the results AFTER the LUT has been applied to the Mini - as that is what the user will see on his screen...

And yes, as posted above, obviously the interpolated points are the ones that need to be validated as only that shows how good you color engine truly is...

is there any way I can define a CUSTOM set of color points in CM (maybe in the color checker section) to read and verify (independent of the LUT box being used) ?

Yes for the radiance we read, calculate the adjustments, write the adjustments and then reread the points. So at the end of the calibration the reads are of the final data we've written.

There are no interpolated points in the radiance, but if you have access to a higher density device, then you can investigate that. We will have some hard data on this soon as well.

In the latest update we substantially updated the list of points you can manually select in video levels.


Edited by sotti - 1/27/13 at 11:35am

Have you done this with CalMAN 5.1 build 1092 or greater. I think you will find we are.

We do in CalMAN 5.1 build 1092 to account for the small but non-linear procsssing going on in the Radiance once the 3D LUT has been applied. In CalMAN 5.1 we have the ability to generate a 3D LUT in our VirtualCube, in hardware for non-linear devices or hybrid if needed.

it would be very helpful to be able to select any point in the gamut for validation, or to be able to import a custom set of validation points... I don't need to check on points that were read ( not even interpolated points), but since the gamut exists of way more points than 5^3 or 65^3 it'll be very interesting to validate random points throughout the gamut, as they will be present in the image and that is what a user sees...

that will tell how good your inner cube is...

Open a 4-Saturation Workflow and change it to 5-Step Saturations from the menu.. You can't get more that 5-Saturation + 20-Step Luminance + Color Checker report with current CalMAN... No way to validate a larger cube with charts of dE etc....

That's what I meant ... 100% agree !

I use Colorchecker (recently allied to Color Comparator).

I don't think these are calibrated points are they?

Incidentally I don't use Lumagen either (Duo) but find the 24 patterns useful just the same.
Edited by PE06MCG - 1/27/13 at 12:51pm

We'll you can look at it with a radiance, but it's not our processing for a radiance, it's the radiance that does the inbetween points. It would be the same for lightspace.

But you absolutely can measure those, all the way up to a 65^3 in CalMAN 5.1

Personally I like to just do a quick run with the colorchecker chart, it's only 24 points, but none of them line up with any of the control points (except the grayscale points), so it's looking at our interpolation.

But you are correct we need the higher density data to prove we are doing it correct.

I want to measure a 10-Point Saturation, how is this possible? Can you give me some details? The Saturations Sweep max value is 5...

To go from 5 steps (20% increments), to 10 steps (10% increments), would only require adding a line or two to our config file.

We can look at adding it to our next release. For some colors like red, that could be useful, for other colors like cyan, it would be complete overkill.

Ok, add 17-Point also, if it's so simple. But again it's something that will be added, currently is not supported by CalMAN.

What are you really trying to see?

Do you really want linear x,y spaced points, or do you really just want to be able to look at RGB spaced data laid out in a more comprehensible way?

dE + CIE Chart Presentations of 10-17 Step Saturation.

Yes but are you looking for linear x,y spacing or RGB linear spacing (so it lines up with the control points of the cube)?

What is the current spacing for 4-5 Step Saturations of CalMAN ?

currently it's a linear x,y spacing, which make for a pretty chart, lines up with the few saturation patterns that were out there (AVS709, Radiance fixed patterns), but doesn't really line up with anything else.

If you wanted to have better perceptual match for distance from white, spacing the points in LAB might make sense. For evaluating just the cube calibration portion using RGB spacing might make sense as the points would line up exactly with the control points.

If you add 10-Point Saturation, from RGB Triplet Target i can create the Disk Chapter for Verification using a Disk, or by using a Pattern Generator like Lumagen.

How do you Verify a larger cube from 5-Point Cube Using CalMAN (Exept Color Checker)?

For example, I want a report from a 16-Point Cube LUT Generated by CalMAN, how can i saw to someone the advandage/difference of a 16-Point Cube vs. 5-Point Cube?

In the more recent 5.1 builds there are a host of static cube sizes you can set via our standard workflow options.

see this post http://www.avsforum.com/t/1453745/calman-5-c6-radiance-mini-3d-calibration-help-please/30#post_22886693

Hi Ted,

Forgive my lack of technical knowledge but I note that you acknowledge Color Checker is one method of verifying the Cube.
Why is it necessary to improve on this?

When you have perform a large 3D-Cube Calibration (4913 Colors), the Color Checker has only 24 Colors Points to measure, 5-Step Saturation has 30 Color Points, i need more Points to check if the performace of a Large LUT, what's why i asked for more Steps of Saturation.

I have measured one 10-Point Cube LUT (1000 Colors) and one 17-Point Cube (4913 Colors) and the Charts of 5-Step Saturation between them are the same. But the actual difference can be seen by swapping this memories with my eeColor Remote. The difference exists but it doesn't exist at Charts.....

Thanks Ted,

I ask because I note that TV camaras and TV monitors may still use only the 24 points of the Color Checker as a reference for original shooting.

Of course, it's used for Camera,Printing,Scanner also

CalMAN's Color Checker is a nice feature that i'm using a lot.

I wonder if these documents are out of date?:

http://tech.ebu.ch/docs/tech/tech3237.pdf

http://tech.ebu.ch/docs/tech/tech3325.pdf

Both seem to refer to the 24 point check as being adequate.

I'm sure the race for better and better accuracy will continue but will human eyesight be able to tell the difference?.
My use of Color Comparator alongside Colorchecker and Saturation Sweeps would suggest that at least my eyes are very inefficient.

Great point. Put a video signal on pause and there can be a visual difference, but in motion it is more difficult to "see" it.