Effect of Std video display settings (w/balance, gain etc) on a pure light source?

[MarcoChip]

Effect of Std video display settings (w/balance, gain etc) on a pure light source?

Hello everybody :smile:
I believe I have a good fundamental question, that so far I've failed to find an answer for - explanation usually revolves around typical analogue camera comparisons.

Outline
I'm looking at combustion via a neutral glass lens to fibre optics, outputing to a Logitech C920 HD camera.
The camera is clamped to a bracket with the lens over a hole - at the other side of the hole are the optical fibres.

No other light can enter the system...... so we are looking at pure light (created by combustion).
IE. The colour that is occurring is the colour we wish to display.

I do understand that there may be an issue of 'calibration', but putting this aside for the moment, I'm really first trying to get my head around the software settings, and what they actually do to the display.

Here are the available settings:

• Exposure
• Gain
• Brightness
• Contrast
• Color intensity
• White Balance

It may be that different companies use different methods to achieve what they call 'gain' (for example).
However, apart from the colour intensity slider (when slid hard left it changes the colour)....... the rest seem to darken the colour (changing towards black)....... or lighten the colour (changing towards white).

So the primary question is 'how do these switches effect the colour display'?

Put into a testable context:

Let's say the low fuel combustion is blue (a torch shining through blue plastic works).

1. With white balance hard left, and all the rest at mid point...... we get a nice blue image displayed.
   As the white balance is slid to the right, the image, gets darker, until very little blue is showing.
3. With exposure hard left, and all rest at mid point...... the image is dark blue.
   As it is slid to the right, the image gets whiter, though we can still see some pale blue.
5. Repeat for gain...... its a similar effect, though less dark (left) and less white (right).... almost as if it were the same as exposure, only on a finer scale (only a subjective view).
7. Brightness..... to the left.... we don't get any black... we get a uniform deep blue.
   To the right... the whole display whitens up (typical brightness response).
9. Contrast.... to the left the blue gets brighter...... to the right.... most of the image has turned black.
11. Colour intensity..... to the left... the blue is gone (I can't discern the colour).... to the right.... the blue is more intense.

Put into a live context

This is really the exciting/interesting part (and I'll have more test data tomorrow)

Low fuel combustion is blue.
Increased fuel combustion is still blue - the engine is producing more power, but combustion is still blue.

As power increases further... combustion turns white.

This could be because:

a) The fuel mixture is too lean (at that point).
b) The exposure or gain (or any of the software switches) are incorrectly set for the heat/intensity range being viewed.

Do you see what I mean?
Even with the torch test, we can make the blue turn white, by upping the exposure and gain etc.

Obviously, the question is:
Is the mixture too lean...... or are my software settings incorrect?

AND :twisted:

Which setting(s) should I be playing with, to try to establish the actual burn colour?

It seems to be, a very nice 'real world scenario' linked to the fundamental question of digital signal processing.
I hope you like it.

:smile:

Tomorrow I'll introduce more fuel higher up in the combustion cycle.
This will give us some hard data..... but it can't confirm as to whether I am pandering to the whims of the software interpretation of the combustion colours that are actually occurring..... but we might learn something :wink:

 

[MarcoChip]

The question begins to focus on:

Exposure & Gain.

There is published opinion that Exposure......over or under, will leave the resultant video without any underlying data that might be retrieved (post recording).

In effect........ if exposure is slid to the left, the low RPM signal that is displayed as black, remains black.......... the colour cannot be reclaimed.

Gain..... it is said...... acts as an amplifier...... increasing or decreasing the voltage....... the implication being that the original information is not lost.

If this is the case, then adjustment should only be made using Gain.

Any thoughts?

 

[BradtheRad]

1.

In dim lighting, gamma is a crucial adjustment. Gamma is not the same as brightness. Gamma particularly lightens the darkest areas of the image. It expands primarily the black-to-grey scale.

Almost all photographs I've taken needed a touch of gamma.

2.

I once read that flesh tones are a good test of color filtering. They are very difficult to get right if they weren't right to begin with.

Therefore if you place a photograph in the same field of view as the flame, with natural looking flesh tones, and you adjust your filtering to make the flesh tones look right, then other colors will be correct too.

 

[MarcoChip]

1.

In dim lighting, gamma is a crucial adjustment. Gamma is not the same as brightness. Gamma particularly lightens the darkest areas of the image. It expands primarily the black-to-grey scale.

Almost all photographs I've taken needed a touch of gamma.

2.

I once read that flesh tones are a good test of color filtering. They are very difficult to get right if they weren't right to begin with.

Therefore if you place a photograph in the same field of view as the flame, with natural looking flesh tones, and you adjust your filtering to make the flesh tones look right, then other colors will be correct too.

That's an interesting concept..... a form of calibration.

RE point 1 - Gamma:
I tried adding post-production gamma in VSDC editor, however in this application it did not produce the desired effect.
Adding gamma did not appear to transform the barely visible colour, to fully visible.
This could be my fault because VSDc is quite a complex editorial suite.
It's still a bit foggy, but understanding the potential effects of gamma helps blow away a bit of the mist.

Lets look at what I did:
I increased the 'fuel to air' ratio, ensuring I had a yellow flame.

At this moment in time.... to record the colour across the range of intensity, requires two passes at different settings.
A high light level, and a low light level.

Because I knew the flame colour was yellow, I was able to judge the video response.

The key control tested (during recording) was 'gain'.
I tried adjusting 'white balance' but it adjusted the colour - so it's eliminated.

The conclusion I reached was and is, apparently obvious..... the next tests should prove it.
The key is going to be 'getting the exposure correct'.
Adding 'gain' was not enough.

However the big question still remaining is:
Does the pixel data contain a colour reference........ even if it is currently being displayed as black, by the video re-play software?

Perhaps I must accept that two passes are required: one for low intensity, and one for high.

The C920 does come with auto balance software, but it is not setup for this application, as it allows just too much intensity.
I may look to see if the parameters can be adjusted, to lower the maximum intensity.

But what is that (actually)?

I don't want the colour being so intense that it becomes white....
however...
... it's not the colour white that is the problem........ it's if a yellow or a blue, is being displayed as white.

Let's say the fuel air mix is 'lean'....... in this case the colour would be whitish, but not brought about by 'over intensity' of a colour.
I guess that's the next test...... a lean mix - which is hot.... so it is intense (what is intense?).

It's got me wondering now, how this will record - low intensity white, and high intensity white.
I can do the physical testing, but I think we're missing the input of a DSP guy, who can explain precisely what is happening when the colour turns white.

The fact is, so far....... to record a good colour of low power combustion......... the setting would be then be too high for recording the colour of high power combustion.

I guess I need to first stick with the 'rich' setting, and try to find an exposure that works for high power, and low power.
Then change to a 'lean' setup, and see if the same settings work....... whilst investigating the question of 'whiteness' in the flame colour.

 

[BradtheRad]

A spectroscope could be useful. It spreads out a light source into a spectrum of individual colors.

A diffraction grating can be a spectroscope. Looks like clear glass, or clear plastic celluloid. A science items supplier (such as Edmund Scientific) may carry an inexpensive version.

Looking through the grating at different light sources, you see the color signatures stretched into a rainbow to both the right and left.

To photograph a spectrum, you would aim your camera to one side of the light source.

The light source will need to be bright. A blue flame is dim to begin with, hence its spectrogram will be even dimmer in the grating.