Hi,
Just throwing this out there for discussion thoughts, I have the ASI2600MC and the sensor amp glow, etc is for all intents and purposes non existent in the histogram for exposures I have looked at 15s - 600s darks the mean figures are around 100 so negligible.
The reason behind the question is that I have started using the brain function on the smart histogram and as such will be getting varying exposure / gain / black levels set by the algorithm involved.
So I would always take Bias / Flats as these will be affected by the gain / black level the debate is whether I really need darks and dark flats, for darks my lights are dithered at a minimum of every 5 frames so I am assuming that any hot/cold pixels in the light frames are going to be calibrated out by Astro Pixel Processor as the image will move but the hot / cold pixels will stay stationary.
For dark flats this is probably a little more subjective as there is no dither in these however would the bias frame be recording some of this information on the hot/cold pixel or is it purely reflecting read noise on the cell or would hot cold pixels really have a dramatic effect on a flat. In some ways flat darks would not be the end of the world if I need to take them as they are normally in the region of 300ms - 500ms or sometimes faster.
Anyone have any thoughts on this.
Cheers
Nick
ASI2600MC To take Darks / Dark Flats or Not
Re: ASI2600MC To take Darks / Dark Flats or Not
Nick
Real world - this image was captured with an ASI2600MC + Sigma f/1.4 ART lens stopped down to f/2, captured under Bortle 3 skies using unity gain. In keeping with the current thinking for modern low read noise CMOS cameras, no bias frames were used to process this data. Astro Pixel Processor did complain about the lack of bias frames but did continue with the calibration. The JPG does lose some detail compared to the significantly larger PNG image. I am looking forward to the Star Reducer Tool in the next version of APP which will allow me to tame the stars in this image.
With an older CMOS sensor, such as the 183C, I would use the traditional darks/flats/bias calibration frames. I have a sneaking feeling that getting too clever with calibration frames can lead to a detrimental effect on the captured data but lack the knowledge to articulate this in a clear manner.
Dave
To answer that I would process an image with and without the calibration frames, perform a side-by-side comparison and post back to the forum.the debate is whether I really need darks and dark flats
Real world - this image was captured with an ASI2600MC + Sigma f/1.4 ART lens stopped down to f/2, captured under Bortle 3 skies using unity gain. In keeping with the current thinking for modern low read noise CMOS cameras, no bias frames were used to process this data. Astro Pixel Processor did complain about the lack of bias frames but did continue with the calibration. The JPG does lose some detail compared to the significantly larger PNG image. I am looking forward to the Star Reducer Tool in the next version of APP which will allow me to tame the stars in this image.
With an older CMOS sensor, such as the 183C, I would use the traditional darks/flats/bias calibration frames. I have a sneaking feeling that getting too clever with calibration frames can lead to a detrimental effect on the captured data but lack the knowledge to articulate this in a clear manner.
- Orion-ASI2600_60x120s_30-darks_100-flats.jpg (286.24 KiB) Viewed 1632 times
Dave
-
PetePeterson
- Posts: 38
- Joined: Fri Oct 30, 2020 8:39 pm
- Location: Rhode Island, USA
- Contact:
Re: ASI2600MC To take Darks / Dark Flats or Not
Hi Nick,
Not needing darks and flats simplifies life, and it greatly influenced my purchase the ASI2600MC-P. It immediately became obvious that darks were needed. And after a month of use dust motes started appearing, and cleaning never gets them all. So for me, flats are needed as well. And since SharpCap has the option to integrate bias frames with the flats it's a no brainer.
Pete
Not needing darks and flats simplifies life, and it greatly influenced my purchase the ASI2600MC-P. It immediately became obvious that darks were needed. And after a month of use dust motes started appearing, and cleaning never gets them all. So for me, flats are needed as well. And since SharpCap has the option to integrate bias frames with the flats it's a no brainer.
Pete
Re: ASI2600MC To take Darks / Dark Flats or Not
Hi Dave / Pete,
My intention was to shoot Flats as they not only correct for dust mites but also for any illumination discrepancies in the scope / cell so they will always be a requirement. As Dave has stated I think I will run the same data with and without Darks / Dark flats to do a direct comparison between the data after integration as this will be a practical method of assessing the effect.
Pete I do not actually use the live stack in Sharpcap other than as a means to see what I am capturing and how it is looking as I collect all the raw frames and reprocess these into an image via Astro Pixel Processor after the session is complete. I normally always used predefined times / gains and black levels that I already had calibration data for. This time however in the image I still need to process I used the brain function in the histogram to calculate the time I did set it to 30s minimum which is what it chose for my conditions and in the session on Wednesday this gives 360 frames for the 3hr session I then had to shoot 100 darks at 30s seconds each if the brain had come back and said 1min then this would have been even longer. This is the main reason that I am looking at whether the light frame equivalent darks are actually necessary as the brain will play with gain / exposure and black level so until you run it you do not know what the calibration data will be.
Cheers
Nick
My intention was to shoot Flats as they not only correct for dust mites but also for any illumination discrepancies in the scope / cell so they will always be a requirement. As Dave has stated I think I will run the same data with and without Darks / Dark flats to do a direct comparison between the data after integration as this will be a practical method of assessing the effect.
Pete I do not actually use the live stack in Sharpcap other than as a means to see what I am capturing and how it is looking as I collect all the raw frames and reprocess these into an image via Astro Pixel Processor after the session is complete. I normally always used predefined times / gains and black levels that I already had calibration data for. This time however in the image I still need to process I used the brain function in the histogram to calculate the time I did set it to 30s minimum which is what it chose for my conditions and in the session on Wednesday this gives 360 frames for the 3hr session I then had to shoot 100 darks at 30s seconds each if the brain had come back and said 1min then this would have been even longer. This is the main reason that I am looking at whether the light frame equivalent darks are actually necessary as the brain will play with gain / exposure and black level so until you run it you do not know what the calibration data will be.
Cheers
Nick
-
admin
- Site Admin
- Posts: 13330
- Joined: Sat Feb 11, 2017 3:52 pm
- Location: Vale of the White Horse, UK
- Contact:
Re: ASI2600MC To take Darks / Dark Flats or Not
Hi Nick,
my take is that if you are using flat frames then you should subtract either a dark flat or a bias frame from the flat frames when making the master flat. This subtracts out the black level pixel value of the camera meaning that the remaining signal is purely due to the elimination during the flat frame. If you don't do this then the flat correction amount won't be quite correct – let me give an example:
Suppose the flat is exposed so that the peak in the centre of the frame is 70% on the histogram, but due to vignetting the corners are 50%. Suppose also that the black level pixel value is at 2% of maximum histogram.
The correct flat correction for the corners is 1.4167x (68/48) - since the amount of signal measured in the centre due to flat illumination is 68% and the amount measured in the corners due to the elimination is 48%. If you don't take account of the background level by using either dark flat or bias frames, you will instead get a correction amount calculated of 1.400x (70/50), meaning that the flat frame will be under correcting. The difference – only 1% of the total correction also – may not seem like much, but may become apparent in the final image depending on how much stretch you apply in processing. The amount of error will also increase as the amount of work the flat frame is doing to correct for uneven brightness increases – for instance if the corners vignette to 20% of histogram then the correct correction value is 3.777x, but without allowing for the background you would get 3.5x.
If your flat frames are short exposure (less than one second or so) then the difference between bias frames and true dark flats is probably negligible.
Cheers, Robin
my take is that if you are using flat frames then you should subtract either a dark flat or a bias frame from the flat frames when making the master flat. This subtracts out the black level pixel value of the camera meaning that the remaining signal is purely due to the elimination during the flat frame. If you don't do this then the flat correction amount won't be quite correct – let me give an example:
Suppose the flat is exposed so that the peak in the centre of the frame is 70% on the histogram, but due to vignetting the corners are 50%. Suppose also that the black level pixel value is at 2% of maximum histogram.
The correct flat correction for the corners is 1.4167x (68/48) - since the amount of signal measured in the centre due to flat illumination is 68% and the amount measured in the corners due to the elimination is 48%. If you don't take account of the background level by using either dark flat or bias frames, you will instead get a correction amount calculated of 1.400x (70/50), meaning that the flat frame will be under correcting. The difference – only 1% of the total correction also – may not seem like much, but may become apparent in the final image depending on how much stretch you apply in processing. The amount of error will also increase as the amount of work the flat frame is doing to correct for uneven brightness increases – for instance if the corners vignette to 20% of histogram then the correct correction value is 3.777x, but without allowing for the background you would get 3.5x.
If your flat frames are short exposure (less than one second or so) then the difference between bias frames and true dark flats is probably negligible.
Cheers, Robin
Re: ASI2600MC To take Darks / Dark Flats or Not
Right been doing some testing using the Horsehead and Flame image from the other day that I posted in the gallery.
The comparison images below are all processed through APP today with the exact same settings and are the integration result at the point that the integration first completes so no further adjustment in Astro Pixel Processor or Affinity photo. The one has a full compliment of calibration frames so Darks, Flats, Dark Flats and Bias the second only has Flats and Bias frames.
The comparisons are a cropped area at 100% the jpg has been save at 95% quality and the only difference that I can see is that the image with darks and dark flats appears to be very slightly and this is very slightly brighter but to all intents and purposes this will make no difference after post processing etc so as far as I can see there is no benefit to the darks and dark flats with this sensor assuming well dithered data and flat + bias calibration frames.
Not sure if anyone else has any thoughts on this.
comparison 1 - Horsehead by Nick Davis, on Flickr
Comparison 2 - Flame Nebula by Nick Davis, on Flickr
Comparison 3 - Stars nebulosity by Nick Davis, on Flickr
Comparison 4 - Nebula by Nick Davis, on Flickr
Cheers
Nick
The comparison images below are all processed through APP today with the exact same settings and are the integration result at the point that the integration first completes so no further adjustment in Astro Pixel Processor or Affinity photo. The one has a full compliment of calibration frames so Darks, Flats, Dark Flats and Bias the second only has Flats and Bias frames.
The comparisons are a cropped area at 100% the jpg has been save at 95% quality and the only difference that I can see is that the image with darks and dark flats appears to be very slightly and this is very slightly brighter but to all intents and purposes this will make no difference after post processing etc so as far as I can see there is no benefit to the darks and dark flats with this sensor assuming well dithered data and flat + bias calibration frames.
Not sure if anyone else has any thoughts on this.
comparison 1 - Horsehead by Nick Davis, on FlickrComparison 2 - Flame Nebula by Nick Davis, on FlickrComparison 3 - Stars nebulosity by Nick Davis, on FlickrComparison 4 - Nebula by Nick Davis, on FlickrCheers
Nick
Re: ASI2600MC To take Darks / Dark Flats or Not
Nick
Try darks + flats and no bias frames.
Dave
Try darks + flats and no bias frames.
Dave
Re: ASI2600MC To take Darks / Dark Flats or Not
Hi Dave,
My whole plan is that I do not want to have to shoot darks at all as they will be different every time I use the Brain option in the smart histogram.
Cheers
Nick
My whole plan is that I do not want to have to shoot darks at all as they will be different every time I use the Brain option in the smart histogram.
Cheers
Nick
Re: ASI2600MC To take Darks / Dark Flats or Not
Nick,
Run the Brain on a portion of the sky at or near the zenith. Just once. That is your minimum exposure; use it for all objects. Repeat once per year as light pollution grows worse. The Brain's recommendation will be on the "longish" side, but it is always better to be long than too short. Darks can last a year as long as your set-point is a reasonable value like -10C. If you set it lower then your cooler may struggle in the summer heat. In that case redo darks twice per year: winter and summer.
Brian
Run the Brain on a portion of the sky at or near the zenith. Just once. That is your minimum exposure; use it for all objects. Repeat once per year as light pollution grows worse. The Brain's recommendation will be on the "longish" side, but it is always better to be long than too short. Darks can last a year as long as your set-point is a reasonable value like -10C. If you set it lower then your cooler may struggle in the summer heat. In that case redo darks twice per year: winter and summer.
Brian
Re: ASI2600MC To take Darks / Dark Flats or Not
Nick,
You can use one exposure for all objects, or you can sort objects into two classes: medium-to-bright and faint-to-medium. The objective is to choose an exposure that ensures a decent signal-to-noise ratio per frame. The reality is that if your SNR is too low (say less than 1.0) then you are constantly trying to dig out of a hole with each frame. You can do it, but your total integration time becomes unreasonably long. It is best to increase exposure until SNR is greater than 1.0. There is no such thing as a perfect exposure, but there is a minimum as measured by the Brain -- don't go below it. I suggest using the Brain's recommended value for medium-to-bright objects such as M81/M82/M51. For fainter objects like M33, double the exposure or whatever you feel comfortable with. However, don't feel compelled to create more than two exposure classes. Two is just fine.
Brian
You can use one exposure for all objects, or you can sort objects into two classes: medium-to-bright and faint-to-medium. The objective is to choose an exposure that ensures a decent signal-to-noise ratio per frame. The reality is that if your SNR is too low (say less than 1.0) then you are constantly trying to dig out of a hole with each frame. You can do it, but your total integration time becomes unreasonably long. It is best to increase exposure until SNR is greater than 1.0. There is no such thing as a perfect exposure, but there is a minimum as measured by the Brain -- don't go below it. I suggest using the Brain's recommended value for medium-to-bright objects such as M81/M82/M51. For fainter objects like M33, double the exposure or whatever you feel comfortable with. However, don't feel compelled to create more than two exposure classes. Two is just fine.
Brian