monochrome DSLR and histogram center

[dts350z]

With a monochrome DLSR, even in L vs. RGB, the histogram displays two peaks. Taller and lower toward black (apparently what would normally be the green pixels) and shorter and more towards the white (apparently what would normally be the red and blue pixels).

 

The first peak is twice as high because there would normally be 2x green pixels compared to red and blue? The first peak more towards the black because  dcraw  is dividing the value by 2?

 

Anyway, my question is for "exposing to the right" should I ignore the first peak?

 

Could we have a switch to make the histogram truly mono?

 

 

[s3igell]

Remember: You seem to have the ONLY "Mono'd Canon DSLR" amongst anyone active on the BackyardEOS Forum.  The rest of us can only Guess at how the Canon Firmware will treat the Sensor Output without the standard Bayer Matrix.  (And it is the Canon Firmware which is producing the Embedded JPG data In-Camera, which eventually leads to the Histogram that BYE produces and displays.)

 

That being said, I don't believe that the Green Pixels are being interpretted directly as Black or the Red&Blue Pixels as White.  That makes little sense, as this no such thing as "Black Light" - and the Electronics of CMOS Pixels can't directly detect "Black".  Instead, all Pixels can detect is the "Amount" of Photons hitting the Pixel.  The rest of it is a product of the De-Bayering Routine (which is still applied for the JPG data, even if there was no Physical Bayer Filter).

 

But, it would be interesting to see a Screenshot or two...

[astroman133]

Normally, the histogram's Y-axis would be normalized so that the green data would not be twice as high as the red and blue. I would suggest that double hump of your image is due to the nature of the image, in that it apparently has 2 clusters of brightness values. As I understand it, BYE is calculating the histogram from the JPG data. And because the camera does not know that the Bayer filters have been removed, the displayed histogram would not be particularly useful.

 

I would take one of your RAW mono images and look at the histogram with some software that treats it as a monochrome image. That means that you would need to convert the RAW file to FITS without Debayering it.

[admin]

Even if you have a mono DSLR it still thinks it is a color camera and its firmware still sees the data as LRGB channels.

[dts350z]

Nikon d5100, not canon.

 

Screen shots to follow.

[astroman133]

Oh. You should probably have posted in the BYN forum.

 

I don't believe that Guylain has said how the histogram is calculated for Nikon cameras. Does the BYN histogram compare to the histogram that is displayed by the camera? If so, then they are both using the JPG image data, which has been de-bayered and stretched. It will not be particularly useful for a mono-modified camera.

 

Not sure what screen shots will show that you haven't already explained.

[dts350z]

Here you can see the double hump, both in "L", and "RGB"

 

Note that in "RGB" 100% of the first hump is "Green".

 

Every picture is like this, regardless of content.

 

I understand that mono DSLRs are rare and if support is limited I understand, however I would like help in understanding how to best use BYN in that case.

[dts350z]

dcraw -D ?

[admin]

The histogram is the same in both BYE/BYN.

 

It is based on the loaded images (as displayed).

 

For a mono DSLR the image still contains RGB channels data and it is represented as such.  The L channel is simply a greyscale representation of the RGB channels.

[admin]

Here you can see the double hump, both in "L", and "RGB"

 

Note that in "RGB" 100% of the first hump is "Green".

 

Every picture is like this, regardless of content.

 

I understand that mono DSLRs are rare and if support is limited I understand, however I would like help in understanding how to best use BYN in that case.

 

 

Open the image with paint.net or any other software, the histogram will be similar if not identical.

 

You see 2 humps because this is how the mono image is being represented based on the actual RGB channels data.  The image is still a color image and thus each channels has data in it and that is what is being represented by the histogram.  The humps are representing data from each channels individually.

 

Hope this helps.

[astroman133]

One of your other questions was regarding how to use the histogram to adjust your exposure. I would not ignore/disregard either of the peaks, but instead make sure that ALL your data is well away from the left and right sides. Beyond that, I would expect that the BYN histogram is of limited usefulness, due to the modifications made to the camera.

 

It may be useful for you to post one of your NEF images to DropBox so others could use it to examine the histogram of the monochrome data with software.

[s3igell]

OK, here's a "Theory":

The In-Camera Debayer Process, which we've established is the 1st part of the Ultimate Origin of these Histograms, is Calibrated for a GRBG (or RGGB or whatever) Pattern with 2x Green Pixels per any "Set of Pixels" by using a Offset Multiplier to give the resulting Image a Natural White Balance.  It is the fact that the Removal of the Bayer Mask allows ALL Pixels to collect the entire Visible Spectrum, combined with the Offset Multiplier raising the Red and Blue Values (or suppressing the Green Values), that is producing the unique Pattern of Dual Humps.  (You could confirm the Constituents of the two Humps by taking an Image or two with a Colored Filter of any one R/G/or B.)

[dts350z]

So I think if we treat the histogram as a black box, whatever it is doing, as long as I expose so there's some space on the left of the lowest hump, I should be fine.

 

As a side note, finding software that understands and will support this corner case is a struggle. Deep Sky Stacker is going to insist on DeBeyring, DeMosiacing, or whatever you want to call it, any RAW DSLR file, while in CCDStack2 you can tell it that the camera is NOT one shot color, and it will treat it as monochrome.

 

Going straight into photoshop/camera raw is also not an option, you need to convert raw files first to tiff or something, by a program that does let you avoid Beyer processing.