Blue Color Take over the picture

[Asaf Benatia]

Here is the sample:

Vimeo

what can i do to fix it?
what can i do to prevent it?

many thanks

[Chris Soucy]

First off, I can't answer you questions definatively.

You would help your cause hugely by taking that clip and compressing it till the pips squeaked.

It is, here at least, the download from Hell and you need all the help that you can get with this one.

I see from your double post that you have a JVC something or other - CMOS, I presume.

Now, as to the cause of this.

The following is pure conjecture as I DON"T KNOW, never seen anything like it.

However, best guess(es):

1. You camera is set to be particularly sensitive to blue, doesn't work for me as the rest of the colours(considering the disco lighting) seem fine.

2. Mixed in amongst the blue strobes are a good few UV sources as well, which are being picked up by the camera as blue, ergo, total blue overload. This one really rings my chimes as I've never heard of a UV filter on either a CCD or CMOS sensor. IR yes, UV, no.

If you could, with the naked eye, see the tell tale flourescing caused by UV lights, my suspicions will be confirmed.

If not, hmm, this is really getting hard.

Have a go at compressing that clip to make it a bit more user friendly and see if anyone else wades in, I'd love to have an answer too.


CS

[Asaf Benatia]

Hi,
Thanks for the answer,
The camera i'm using is JVC HM700 (CCD sensor)
The light is Blue LED light, very close to the subjects .

[Chris Soucy]

Are you sure?

Must have better LED's there than I've ever seen, heck, you'd be lucky to light a letter box with any that are available here (tho' guess that depends on how many mega bucks one wishes to spend).

No UV's?

Don't know, is the answer, can't figure out how blue would take over an entire frame, even if it is the primary light source.

Shall sleep on it.


CS

[Dave Blackhurst]

OK, I'm going the opposite direction - that clip is so badly pixellated and so low resolution that it's very difficult to sort out what is overcompression/bad compression and a possible abnormal/malfunction.

Full frame captures at HD resolution might help.

That said, possible theories:

Whatever compression you used is somehow over-reacting to blue... but I'll presume the original footage has issues as well... leading to theory #2...


EDIT: got this confused with the new PANASONIC "700" series cams... if this is a JVC HM700, it's CCD, not CMOS... but it is supposed to be HD... the footage looks like SD to me... and the pixellization was very reminiscent of certain Panasonics I've used...

Either way, you've got three sensors, one each for Red, BLUE, and Green - perhaps the blue sensor is "hot" compared to the other two - this reminded me initially of how it looks when one CCD started to go on older cameras (and Panasonic had a few issues there IIRC leading to recalls to replace CCD blocks). I did see pretty strong frame wide color shifts as red and green lights flashed, it just looks like the blue was more agressive, and caused adjoining pixels to bleed blue, almost like a flash going off. There were still sections of the "bluewn" frames that looked fine though...

One last possibility - what WB settings were you using? I wonder whether the cam was in "auto" and kept trying to adjust WB, resulting in completely skewed colors as the cams "brain" attempted to re-WB with each laser color change? In difficult light conditions you're better off to use a preset and fix in post.

Not being familiar with JVC's, I've got one other "WAG" - that blue spectrum light is probably "darker" relatively to the other R/G light - it occurs to me that in the imaging DSP perhaps the individual sensors have separate gains under auto control, and when the blue sensor is interpreted to "need" more "gain", the cameras processor "cranks" that channel in an ill fated attempt to maintain "balance" in the brightness of the overall image? I notice that the portions of the frames that look "normal" seem to be have very distinct edges, almost like you'd see with blue/greenscreen footage (keeping in mind the footage is awfully low rez, so this may not hold true if we can have some high res clips to look at). It almost appears to me that the camera is processing the regions of the image independently, then creating the frame from the interpreted bits, thus this strange hypothesis. "Digital Signal Processing" can sometimes consist of some very complex approaches that may malfunction under unexpected conditions, when it turns out the camera isn't as "smart" as it was supposed to be.

[Chris Soucy]

CCD it is, doesn't invalidate anything you've said (I don't think) just to keep the record straight.


CS

[Thomas Klausnitz]

Hi Asaf,

the reason is the gamma intensity of blue in the color spectrum.

The next one is the technical one, the ccd´s are build with silicium or other coatings. The manufactures uses different types of them, silicium, as far polysilicium and indium-stannic-oxyd.

The new generation of ccd´s build with indium-stannic-oxid coating, this has less optical attenuation in light spectrum of 400 to 540nm. The result of this is a higher blue sensitivity. In this case, on ccd´s came 3 times more of blue light, for ab better signal-noise-ratio to get a better color dynamic.

The only one to prevent it, use less blue light in proportion to green and red. (blue: one out of three or half one). Make your adjustments in blue color stage light (by checking the reflection on white).

By fixing the problem in post production you will be get some problems because the record is overlightet by blue parts and i´ve no idea how you get the blue overlight out.
You have a composed record, complete with all information in videostream and it´s not possible to extract the blue light from video or adjust the gamma down for blue, and if you do that you get an horrible result. Which informations won´t show in the blue area of record, theres no other picture pieces as blue, black.

I´m sorry, but my opinion is, you can only fix the problem by prevent to much blue light.

Greets,

Thomas

[Dave Blackhurst]

Chris -

Hadn't seen the cross post, and when I did, I realized it was posted under the JVC HM700 section... not familiar with that cam, but looked it up, and yup, CCDs... which does raise some questions of CCD "smear".

I've edited my earlier post a bit to reflect this, and added one other wild guess...

[Thomas Klausnitz]

Red Color cames sometimes also with this phenomena.

[Asaf Benatia]

Hi
yes it is HD , but downconversion upload.
i will try to upload some stills captures.

[Asaf Benatia]

Here is the pic

[Chris Soucy]

Yep, that's impressive.

As I said earlier, I might sleep on it, 'cos it's got me stumped, unless there's a serious problem with either the camera or the encoding.

What, exactly, did this look like in the viewfinder/ LCD when you were shooting it?

The same, or didn't notice?

With that I'm signing off, it may only be 2.14 pm where you are, it's 11. 47 pm ( surely not) here.

Back to the fray tomorrow.


CS

[Asaf Benatia]

looks the same on the lcd

[Thomas Klausnitz]

Blue LED Light has the clothest light spectrum as real daylight.
The problem by recording on direct stage lights, spots is that various colors has different wavelenght.

blue, and this is physicaly base, has a higher light energy than red or green.

the technical base, in gy-hm700 is an CCD and this part converts the light in juice.

the reflections in the screenshot show you the intensity of this led stage spots by reflecting.

i asked a guy by TV and he told me you should avoid direct spotlights in base colors with to much intensity, otherwise you have always problems with overlight screen in one color like the good old tube cams. you can see some effects on live stage records, thats the same one.

[Calvin Bellows]

I have a HM 700 and I was shooting a PFC fight and the same thing happened with the fighter in the blue corner. I don't know if there is a fix or something that just happens with these cameras and very blue light. Saying that when I use tungsten with blue gels it doesn't happen.