Raw images super dark?

I feel stupid asking this but I’m confused. Prior to SGpro I used byeos , now when I look at the RAW images SGpro puts out they are so dark compared to before. Does byeos do an image stretch when it displays the image? Also if I open the raw in windows it looks stretched so does SGpro show the raw image as it actually is? Reason I’m asking is I had a hard time getting my exposure time right. I was at 10 minutes using a DSLR and LP filter and never reached the 1/3 histo

What do you mean by “never reached 1/3 histo”? Are you talking about the peak of the histogram? Do you mean the unstretched histogram? If so don’t look so much at the peak and try the get the max pixels close to or slightly above saturation.

I had been taught to try and get the histogram 1/3rd of the way off of the left. When I used BYEOS it was generally a 300 second exposure at 1600 ISO and I had a good idea of the data I was getting. When looking at the raws that SGpro produces they seem considerably darker and even at 600 seconds my histogram was only about 1/4th of the way. I don’t understand what you mean to get the pixels close to or above saturation

Dont use SGPs histogram for dslr. Download fastone image viewer. Load the image and press control H.

Sgps histogram is better suited for astro cameras.

You can monitor the mean value in image statistics too with SGP instead of histogram once you know what number to watch for.

The 1/3 histogram rule for DSLRs only applied to the back of camera LCD histogram. Due to the way it is written, you can also apply this rule for BackyardEOS/BackyardNikon, as they basically do the same thing as the LCD histogram. These all apply a tone curve (think, DSLR picture style) to the data, which effectively stretches it a bit for you.

For any other program, such as SGP, or PixInsight, etc. your images will be rendered LINEARLY, without any stretching. For DSLR users this can be a bit confusing at first, however this is also how CCD camera data is displayed as well. A linear histogram is usually just a spike that is pretty close to the left edge of the histogram. This is normal and expected of astro photography data, as we work with very weak signals. The key is making sure that separation between the spike and the left edge is large enough, and that depends on the camera and the ISO setting.

I generally offer a simple formula for how to measure your data, in a program like SGP, based on the read noise and the gain. You will need to measure your data for this, and that could be done with PixInsight’s BasicCCDParameters script, or possibly SharpCap (the latter being a lot cheaper). Once you have your read noise (in e-) and gain (in e-/ADU), you can use these formulas to help you figure out, with simple measurements of the background sky, whether you are exposing enough or not:

BGMin = ((r^2 * 3 / g) + o) * 2^16/2^b
BGIdeal = ((r^2 * 10 / g) + 0) * 2^16/2^b


r = read noise (e-)
g = gain (e-/ADU)
o = offset (ADU; Canon = 2048, Nikon = 600 (if no black clipping, otherwise 0))
b = raw file bit depth (usually, 14; rarely, 12)

Generally, expose at least enough to get to BGMin, and if you can, expose enough to get to BGIdeal. Somewhere between those two is acceptable as well, however the lower you go below BGIdeal, the more read noise will impact your results.

For the equation

BGideal= ((r^2 * 10 / g) + 0), (16 bits)

For the offset, Can I consider the mean value from a Bias?

I got a QSI660 and Bias mean values are around 510 ADU with a mínimum = 375 and Max= 64000.



Sure, mean or median of the bias would do.

Unfortunately sharp cap won’t do a sensor analys on dslrs. For now I’ll have to use faststone image viewer until they add that support. Unless anyone knows a way?

So if I did my research correctly for my Canon 700D my R=2.7 and G= about .075. Do I use 2048 as a blanket canon offset and what is the BGmin? Is the min/ideal the mean of the image