Dither Distances Simplified

[Yoddha]

Hi Steve,

Welcome to the forum!

If the formula gives you 8.7 you have to use a bigger dithering distance than 1, try with 7-8. Also try to increase the Settle distance to 1.0

[[email protected]]

For Dave
Currently running PHD2 ver 2.6.11, will look up development version
Currently running ATP ver 4.3, will dowload version 4.31
Let's see if it helps
Thanks

For IVO
I am a bit confused. In reading all the other posts I got the impression that a K value of 8.7 (rounded to 9) would give me an MDD of 9 if I entered "1" in the dithering distance of the guide settings of ATP. If I entered 8 to 9 as the dithering distance I would get an MDD of somewhere between 72 and 81. I assumed 8 to 12 would be reasonable value.
Can you clarify......Thanks

[Drac0]

For IVO
I am a bit confused. In reading all the other posts I got the impression that a K value of 8.7 (rounded to 9) would give me an MDD of 9 if I entered "1" in the dithering distance of the guide settings of ATP. If I entered 8 to 9 as the dithering distance I would get an MDD of somewhere between 72 and 81. I assumed 8 to 12 would be reasonable value.
Can you clarify......Thanks

I'm confused with your reply too Ivo.

I have a spreadsheet setup to work everything out using the formula: K = 2 x (FLi / Flg) x (Pg / Pi), where the final APT dithering distance is the required MDD (in my case 9) divided by K.

I've set this up for my various setups using guide/imaging scopes, cameras & bin settings. It all works out that the higher the K value, the lower the dithering needs to be in APT. I aim for a MDD of at least 9, so if K>9 I use a dithering distance of 1 & if K<9 I use 9/K (rounded up) to get the dithering distance required. Did I misinterpret the information I found?

MDD.png

As for your issue Steve, I would have been more inclined to lean towards it being related to the Dithering Stability you're using, maybe it's too small? There is no real issue with setting the stability higher to 1.0, 1.5 or even more, especially at longer focal lengths where the slightest breeze can move the telescope more pixels than that. I use 1.50 as a default with a 15 second settle time & a timeout of only 30 seconds & rarely have dithering fail, but I'm using 1200mm focal length at most right now.

Cheers,
Mark

[Yoddha]

Hi Steve, Mark,

You are absolutely right! Sorry for introducing confusion

[Drac0]

Revisiting my guiding settings and I have to ask why in the original formula ( K = 2 x (FLi / FLg) x (Pg / Pi) ) is the 2x multiplier used? Shouldn't it just be based on the scale between the guide\imaging combination? In my case 1 pixel of the guide camera = 2.59 pixels on my imaging camera. Why would this need doubling? By doubling it you are actually going to halve the maximum dithering distance and end up nowhere near what you were after. Available online dithering calculators will actually show this, with the required dithering setting to achieve a desired MDD being twice what this formula provides.

Confused...

Cheers,
Mark

[jkcolli]

In an earlier post Steve described how he came up with MDD as follows:

https://aptforum.com/phpbb/viewtopic.php?f=19&t=3154

“ I performed two dithering experiments. In each case I allowed the system to dither about 20 times. I then used DSS to measure the dX and dY dither positions for each sub. I then found the average of the dX values and subtracted that from all dX readings. I did the same for the dY readings. This determines the centroid of the random or spiral ditherings. The distance of each dither from the centroid is what I refer to as the "dither" in the discussion below.

DITHERING:
1) For APT Dither Distance = 5 and PHD2 Dither Scale = 1 the maximum measured dither (best estimate of MDD) was about 10 or 11 guide camera pixels.
2) For APT Dither Distance = 2 and PHD2 Dither Scale = 1 the maximum measured dither (best estimate of MDD) was about 4 guide camera pixels.

So, if the PHD2 Dither Scale = 1.0 it appears that the MDD (maximum dithering distance) in guide camera pixels will be 2 x (APT Dither Distance ) “

So the 2 comes from two experiments that he did.

[Drac0]

In an earlier post Steve described how he came up with MDD as follows:

https://aptforum.com/phpbb/viewtopic.php?f=19&t=3154

So the 2 comes from two experiments that he did.

Yes, I've read that. I've just been going through it again & I think I found here Steve has come up with it & the error he's made. He's done his calculation based solely on his gear. And from that post both the guide & imaging camera have the same pixel size while the focal length of the imaging scope is only just over twice the guide scope. That gives a ratio of only just over 2x for his combination. So his calculations will work - for his setup. I don't think it will work correctly for others that have a different guide:imaging ratios.

And in his first post of this thread he's made an error in his calculation by mixing up the pixel sizes for the cameras. In truth, every one pixel move of his guide camera will only move his imaging camera 1.58 pixels, NOT the 5.98 pixels he thinks, nor the 3.16 pixels his 'corrected' calculation would give. Adding the 2x multiplier just throws everything off, making people think their MDD will be twice what it actually is.

Cheers,
Mark

[Drac0]

For anyone interested here's a dithering distance calculator for APT I threw together. You just need to enter the focal length, pixel size & binning for the imaging & guide setups, along with the number of pixels of the imaging camera you wish to have as your MDD. It will then give you the figure you need to enter in APT to achieve that. It's in open document format so you should be able to open it in any spreadsheet program.

APT Dithering Distance Calc.ods

Cheers,
Mark

[jkcolli]

I don't think Steve333 made any mistakes in his calculations as I have shown below.
Also, I have included a better description of how he came up with the factor of 2.
And I have included the results from similar experiments that I have done.

Steve333's equipment and calculations
Guiding Camera Orion Star Shoot Autoguider Pixel size(Pg) = 5.2micron
Guiding Scope Orion 50mm Guild Scope Focal Length(FLg) = 162mm. (rounded to 160mm by Steve333)

Imaging Camera ASI1600MM Pixel size = 3.78 micron
Imaging Scope William Optics RedCat 71 f/4.9 Focal Length = 162mm


Guiding Scale = 206.265*Pg/Flg = 206.265*5.2/160. = 6.70 arcsec/pixel
Imaging Scale = 206.265*Pi/FLi = 206.265*3.78/348 = 2.24 arcsec/pixel
Guiding Scale/Imaging Scale Ratio= 2.99

With different factoring as done in Steve333 equations:
Guiding Scale/Imaging Scale = {206.625*(Pg/FLg)}/{206.625*(Pi/FLi)}
= (FLi/FLg) x (Pg/Pi)= 348/160) X (5.2/3.78)= 2.99


This means that the movement of one guide pixel is equal to 2.99 imaging pixels.

This agrees with the results of your calculator(2.992).

Steve333's factor of 2
The factor 2 that Steve333 came up with was from doing two experiments looking at the actual distances the guiding camera moved in pixels.

PHD2 Random Walk dithering
When dithering occurs, PHD2 moves a random amount in X from 0 to the APT Dither Distance and at the same time it moves a random amount in Y from 0 to the APT Dither Distance.

On the next dither it moves from the point it just moved to another the next point. This is a random walk in two dimensions.

What Steve333 did was look at two examples of this random walk and found in those two examples that the maximum distance from the centroid of the random walk pattern was approximately equal to two times the APT Dither Distance. That is where the factor of 2 comes from.

Theoretical Random Walk equation
For a 2D random walk with N equal to the number of steps:
Expected absolute distance from the origin = sqrt(2N)* [Gamma Function(3/2)=Sqrt(2N)*0.8862 = 1.25*sqrt(N)

I don't know how important the absolute distance is, but it is clearly greater than the maximum step size in X and Y.

Additional Dither Experiments
I repeated the experiments of Steve333 except I used the actual dither distances in pixels that are reported in the PHD2 log.

I have included a couple of charts with an APT Dither Distance of 2.
You can see due to the random walk the distances moved are clearly greater than the APT Dither Distance of 2. Note also that these diagrams make no assumptions about Equipment used for Guiding or Imaging. The PHD2 random walk is based solely on APT Dither Distance.

So Steve333's factor of 2 had nothing to do with his particular equipment.

I am not sure what factor should be used yet but it will be greater than 1. I intend to collect more data when the weather allows.

Jack

[vidrazor]

A quick question on using the APT Dithering Distance Calc.ods file, does this presume the PHD2 dithering scale is set at 1? I know in the previous PHD this would divide the setting in APT by half, does this version not do that?

Thanks.

[Drac0]

A quick question on using the APT Dithering Distance Calc.ods file, does this presume the PHD2 dithering scale is set at 1? I know in the previous PHD this would divide the setting in APT by half, does this version not do that?

Thanks.

With APT you don't need to change the dithering scale for PHD2 as you specify the max distance in APT, so it should be left at 1. The scaling in PHD2 is a multiplier that is useful if using a program that doesn't allow you to set a specific max dither distance. So not needed for APT.

From the PHD2 manual:

For some applications that do PHD2 dithering, you can't specify the maximum amount directly - you are perhaps limited to choices like small/medium/large and the max dither amounts will have preset values. For that reason, PHD2 has a dither scaling parameter in the 'Global' tab of the Advanced Settings dialog. It is basically a multiplier term that lets you adjust the range of dither amounts that are possible. So a scale factor of 1 doesn't change the preset value at all, a value of 10 multiplies it by 10X, etc. If you're using an app that lets you specify the maximum amount directly (e.g. PHD_Dither), you should leave the dither scale set to 1.0. Otherwise, you can adjust the scale factor if you aren't happy with the overall range of dithering you're getting with one of the small/medium/large choices..

Cheers,
Mark

[vidrazor]

OK cool, thanks so much!

[mhamburg]

Is the scenario that much simpler when using an OAG and an imaging camera? In my case a Celestron 11 HD (2800mm fl.) and a Canon 6D DSLR.
Michael