APPM Image Scale Question and Suggestions


John Jennings
 

I'm really confused about the terminology in this post. I've been using a QHY294C Pro  since it was released. In 11mp mode, A pixel is defined as 4.63um*4.63um. In unlocked 47mp mode, A pixel is defined as 2.314um*2.315um by the manufacturer. I always use these as the Bin x 1 values in plate solver settings (depending on camera driver mode) everywhere just like a monochrome camera. I do understand that underneath the driver at the hardware level, in the 11mp mode, the pixels in "this camera only" (SONY IMX294 BSI CMOS  (Color version)), are Binned x 2. But that's irrelevant at the image level after the driver output.

As I mentioned before, I've had the same issue with these small pixel settings with my Mewlon 300. However, the new QHY Beta drivers are allowing 4x4 color binning on some of the new color cameras. I've not checked this out on my 294C Pro. This will make a big difference in plate solving slow scopes. In general, I do not treat my color cameras any different than my monochrome camera and always use the manufacturers specified pixel size.


Ray Gralak
 

Hi Marty,

1. With a color camera the appropriate x-scale and y-scale factors to enter are twice the individual pixel dimensions.
In my case I should enter 0.52 rather than 0.26 for each X and Y scale. Is that correct?
This assumes the information you provided is correct (that 2x2 binning results in 1 arc-sec/pixel image scale using the ASI294MC with a 300mm aperature F15 scope).

-Ray


-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of mjb87 via groups.io
Sent: Sunday, April 18, 2021 4:26 AM
To: main@ap-gto.groups.io
Subject: Re: [ap-gto] APPM Image Scale Question and Suggestions

[Edited Message Follows]

Thanks. I appreciate the assistance. Here is what I'm taking away so far

1. With a color camera the appropriate x-scale and y-scale factors to enter are twice the individual pixel dimensions.
In my case I should enter 0.52 rather than 0.26 for each X and Y scale. Is that correct? (Not sure I understand why
it would be.)

2. When you later attempt to redo a platesolve on a saved image, and when it reports its estimated actual image
scale, it is reporting the binned scale. In other words, the reason my reported scale was 1.04 is that the actual
unbinned image scale was 0.52 and I was using 2x2 binning. Is that correct? (If so, it might be helpful to modify that
window to emphasize that the reported scale is binned scale.)

3. I may just be asking too much to get good solutions using an f/15 telescope and a camera with small pixels, given
my Bortle 5 skies and with the moon out. One option is to use a camera with a larger pixel size. As evidence of this,
in using a similar camera on my 130mm GTX I still had to do 2x2 binning to get easy solutions. I use a Lodestar X2
as a guide camera on my other setup. It has 8.2x8.4 pixels. I can try that instead of the ASI1600MC-Cool.

4. Other things to try: more tolerance around image scale, less sigma above mean, longer exposures.

Make sense?


Ray Gralak
 

Hi Marcelo,

Ray wrote:
2x2 binning of 0.26 arc-secs/pixel should be 0.52 arc-secs/pixel, not 1.0 arc-secs/pixel.

However, since this is a color camera, if each group of 4-pixels is considered 1 pixel, then you should enter
0.52 arc-secs/pixel in APPM for the X and Y image scale values.
Marcello wrote:
I assume that the same principles apply for the ASI294MM camera, whose default mode is 2x2 binning, right?
What I posted above was just a theory. The ASI294MM, which is monochrome, recently received a firmware upgrade to "unlock" the higher resolution (natively unbinned). So, it may be that in the color camera (ASI294MC), each "logical pixel" consists of four color-filtered native pixels. So, pixel-scale may depend on the camera modes defined by the sensor and implemented by the camera manufacturer.

-Ray

-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of Marcelo Figueroa via groups.io
Sent: Saturday, April 17, 2021 10:03 PM
To: main@ap-gto.groups.io
Subject: Re: [ap-gto] APPM Image Scale Question and Suggestions

On Sat, Apr 17, 2021 at 05:31 PM, Ray Gralak wrote:


2x2 binning of 0.26 arc-secs/pixel should be 0.52 arc-secs/pixel, not 1.0 arc-secs/pixel.

However, since this is a color camera, if each group of 4-pixels is considered 1 pixel, then you should enter
0.52 arc-secs/pixel in APPM for the X and Y image scale values.

I assume that the same principles apply for the ASI294MM camera, whose default mode is 2x2 binning, right?


mjb87@...
 
Edited

Thanks. I appreciate the assistance. Here is what I'm taking away so far

1. With a color camera the appropriate x-scale and y-scale factors to enter are twice the individual pixel dimensions. In my case I should enter 0.52 rather than 0.26 for each X and Y scale. Is that correct?  (Not sure I understand why it would be.)

2. When you later attempt to redo a platesolve on a saved image, and when it reports its estimated actual image scale, it is reporting the binned scale. In other words, the reason my reported scale was 1.04 is that the actual unbinned image scale was 0.52 and I was using 2x2 binning. Is that correct?  (If so, it might be helpful to modify that window to emphasize that the reported scale is binned scale.)

3. I may just be asking too much to get good solutions using an f/15 telescope and a camera with small pixels, given my Bortle 5 skies and with the moon out. One option is to use a camera with a larger pixel size. As evidence of this, in using a similar camera on my 130mm GTX I still had to do 2x2 binning to get easy solutions. I use a Lodestar X2 as a guide camera on my other setup. It has 8.2x8.4 pixels. I can try that instead of the ASI1600MC-Cool.

4. Other things to try: more tolerance around image scale, less sigma above mean, longer exposures.

Make sense?


Marcelo Figueroa
 

On Sat, Apr 17, 2021 at 05:31 PM, Ray Gralak wrote:
2x2 binning of 0.26 arc-secs/pixel should be 0.52 arc-secs/pixel, not 1.0 arc-secs/pixel.

However, since this is a color camera, if each group of 4-pixels is considered 1 pixel, then you should enter 0.52 arc-secs/pixel in APPM for the X and Y image scale values.
I assume that the same principles apply for the ASI294MM camera, whose default mode is 2x2 binning, right?
 


Mike Dodd
 

On 4/17/2021 8:23 PM, Ray Gralak wrote:
What am I missing?
Marty said 2x2 binning was producing images with 1 arc-sec/pixel
scale. What other reason would there be for 2x2 binning to produce 1
arc/sec pixel when each pixel is 0.26 arc-seconds? It *should*
produce 0.52 arc-sec/pixel image, right?
Yes. Something didn't "click." Sorry for my confusion.

--- Mike


Ray Gralak
 

What am I missing?
Marty said 2x2 binning was producing images with 1 arc-sec/pixel scale. What other reason would there be for 2x2 binning to produce 1 arc/sec pixel when each pixel is 0.26 arc-seconds? It *should* produce 0.52 arc-sec/pixel image, right?

-Ray

-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of Mike Dodd
Sent: Saturday, April 17, 2021 3:43 PM
To: main@ap-gto.groups.io
Subject: Re: [ap-gto] APPM Image Scale Question and Suggestions

On 4/17/2021 6:31 PM, Ray Gralak wrote:

However, since this is a color camera, if each group of 4-pixels is
considered 1 pixel, then you should enter 0.52 arc-secs/pixel in APPM
for the X and Y image scale values.
I don't understand that, Ray. A pixel is a pixel regardless of the color
filter over it.

When an image is debayered, the debayering algorithm gets light values
from every pixel, and assigns RGB values to the same pixels in the color
image.

the resulting color image has the same number of pixels as the
non-debayered image. The image size is unchanged.

What am I missing?

--- Mike







John Jennings
 

Mike,

I also have a 130GTX with Quad reducer/flattener on a APMach1. With a full frame QHY410C or my QHY268C (APS-C) it literally screams through the sky plate solving instantly like a maniac in the same Bortle 8 skies.

John


John Jennings
 

Mike,

You didn't mention if you are in a light polluted area or have dark skies. I live in a Bortle 8 sky. I had a lot of trouble Plate Solving with my Mewlon 300 reduced to f10 with the small pixels (4.63micron) of the QHY294C. I see your using USNO A2.0 for the narrow field of view. That helps a lot, but sometimes the GSC would work better in a specific area. For me in the city, it depended on what area of the sky I was solving for and how many stars were available. When I was doing a mapping run with APPC Pro with my AP900, if the Milky Way was overhead, it was a easy. At other times when the sky was star sparse, it was erratic.  I tried all types of binning with inconsistent results.  In the end, I switched  for a test to an old big pixel camera QHY8 (7.8micron) and binned x 2 and never had a problem. Almost 100% solves. I created a perfect mapping run and switched back to the 294C for imaging. When I finally switched to a full frame camera, it became easy too with the small pixels.  It's just tough with a narrow field of view, small pixels, small chip,  slow f ratio and bright skies.


Mike Dodd
 

On 4/17/2021 6:31 PM, Ray Gralak wrote:

However, since this is a color camera, if each group of 4-pixels is
considered 1 pixel, then you should enter 0.52 arc-secs/pixel in APPM
for the X and Y image scale values.
I don't understand that, Ray. A pixel is a pixel regardless of the color filter over it.

When an image is debayered, the debayering algorithm gets light values from every pixel, and assigns RGB values to the same pixels in the color image.

the resulting color image has the same number of pixels as the non-debayered image. The image size is unchanged.

What am I missing?

--- Mike


Ray Gralak
 

Hi Marty,

2x2 binning of 0.26 arc-secs/pixel should be 0.52 arc-secs/pixel, not 1.0 arc-secs/pixel.

However, since this is a color camera, if each group of 4-pixels is considered 1 pixel, then you should enter 0.52 arc-secs/pixel in APPM for the X and Y image scale values.

-Ray

-----Original Message-----
From: main@ap-gto.groups.io [mailto:main@ap-gto.groups.io] On Behalf Of mjb87 via groups.io
Sent: Saturday, April 17, 2021 9:12 AM
To: main@ap-gto.groups.io
Subject: [ap-gto] APPM Image Scale Question and Suggestions

Hi everyone.

I have used APPM successfully with my Mach2 and 130mm Starfire GTX. I am now trying to build a model for my
1100 with a CFF 300 on it. I have some specific questions about image scale. I'm using USNO A2.0.

The camera is a ASI1600MC-Cool with a 0.67 focal reducer on a 300mm f/15 Cassegrain. I calculate unbinned
image scale at just under 0.26, which is of course pretty small. I binned 2x2. My initial run was not successful --
many failures, often reported as due to potential image scale issues. Later, using the image-link-test solve
capabilities, I was able to get a couple of the earlier images to solve, but the estimated image scale in those
calculations was reported just over 1.0. When I re-ran the image-link-test solving with that 1.0 entered for image
scale I got about half the images to solve easily.

Question #1: Is this inconsistency because of the 2x2 binning? Does the plate-solve results in the Image-Link-Test
display estimated unbinned or binned image scale? The math works: 4 x 0.26 is about what was reported.

Question #2: Just to confirm, then, that when I enter X-scale and Y-scale factors in the platesolve setup, I should
enter the unbinned values, right (0.26)? The software will then adjust, I assume, for the binning.

Question #3: What options should I play with to reduce the potential for failed solutions due to image scale issues.
How high can I set (realistically) the Image Scale Tolerance? Is 50% too high? 100%? I also assume it would be
helpful to set Catalog Expansion to a higher value, maybe to the max at 0.8. I'n not trying to get fast solutions, just
more solutions.

Question #4: Any other suggestions for running APPM with such a small image scale? I was worried about getting
enough stars so I increased the exposure time on the camera (from 5 to 15) and reduced the sigma above mean
(from 4 to 3). Anything else?

Thanks.

Marty


mjb87@...
 
Edited

Hi Brian,

I did exactly that. Entered 0.26 for both X and Y and then 2x2 binning. I only raise it because when you look at the results *(when I retested some images the next day) of image-link testing when it reports the image's estimated scale it reported 1.04. So I suspect the latter is binned results. 

Marty


 

Hi Marty

Just to confirm, you should enter the unbinned image scale, and then choose your binning in APPM

I'm not sure what you did in this regard

Brian


On Sat, Apr 17, 2021 at 9:12 AM mjb87 via groups.io <mjb87=verizon.net@groups.io> wrote:
Hi everyone.

I have used APPM successfully with my Mach2 and 130mm Starfire GTX. I am now trying to build a model for my 1100 with a CFF 300 on it.  I have some specific questions about image scale. I'm using USNO A2.0.

The camera is a ASI1600MC-Cool with a 0.67 focal reducer on a 300mm f/15 Cassegrain. I calculate unbinned image scale at just under 0.26, which is of course pretty small. I binned 2x2. My initial run was not successful -- many failures, often reported as due to potential image scale issues. Later, using the image-link-test solve capabilities, I was able to get a couple of the earlier images to solve, but the estimated image scale in those calculations was reported just over 1.0. When I re-ran the image-link-test solving with that 1.0 entered for image scale I got about half the images to solve easily.

Question #1: Is this inconsistency because of the 2x2 binning? Does the plate-solve results in the Image-Link-Test display estimated unbinned or binned image scale?  The math works: 4 x 0.26 is about what was reported. 

Question #2: Just to confirm, then, that when I enter X-scale and Y-scale factors in the platesolve setup, I should enter the unbinned values, right (0.26)? The software will then adjust, I assume, for the binning.

Question #3:  What options should I play with to reduce the potential for failed solutions due to image scale issues.  How high can I set (realistically) the Image Scale Tolerance?  Is 50% too high?  100%?  I also assume it would be helpful to set Catalog Expansion to a higher value, maybe to the max at 0.8.  I'n not trying to get fast solutions, just more solutions.

Question #4: Any other suggestions for running APPM with such a small image scale?  I was worried about getting enough stars so I increased the exposure time on the camera (from 5 to 15) and reduced the sigma above mean (from 4 to 3). Anything else?

Thanks.

 Marty



--
Brian 



Brian Valente


mjb87@...
 

Hi everyone.

I have used APPM successfully with my Mach2 and 130mm Starfire GTX. I am now trying to build a model for my 1100 with a CFF 300 on it.  I have some specific questions about image scale. I'm using USNO A2.0.

The camera is a ASI1600MC-Cool with a 0.67 focal reducer on a 300mm f/15 Cassegrain. I calculate unbinned image scale at just under 0.26, which is of course pretty small. I binned 2x2. My initial run was not successful -- many failures, often reported as due to potential image scale issues. Later, using the image-link-test solve capabilities, I was able to get a couple of the earlier images to solve, but the estimated image scale in those calculations was reported just over 1.0. When I re-ran the image-link-test solving with that 1.0 entered for image scale I got about half the images to solve easily.

Question #1: Is this inconsistency because of the 2x2 binning? Does the plate-solve results in the Image-Link-Test display estimated unbinned or binned image scale?  The math works: 4 x 0.26 is about what was reported. 

Question #2: Just to confirm, then, that when I enter X-scale and Y-scale factors in the platesolve setup, I should enter the unbinned values, right (0.26)? The software will then adjust, I assume, for the binning.

Question #3:  What options should I play with to reduce the potential for failed solutions due to image scale issues.  How high can I set (realistically) the Image Scale Tolerance?  Is 50% too high?  100%?  I also assume it would be helpful to set Catalog Expansion to a higher value, maybe to the max at 0.8.  I'n not trying to get fast solutions, just more solutions.

Question #4: Any other suggestions for running APPM with such a small image scale?  I was worried about getting enough stars so I increased the exposure time on the camera (from 5 to 15) and reduced the sigma above mean (from 4 to 3). Anything else?

Thanks.

 Marty