Register    Login    Forum    Search    FAQ

Board index » Products » Cameras




Post new topic Reply to topic  [ 51 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6  Next
Author Message
 Post Posted: Wed Oct 11, 2017 8:07 am 
Offline
User avatar

Joined: Thu Feb 21, 2013 2:51 am
Posts: 2281
Hi Hiten
it must be the brightness is not enough when you want to bin
so if increasing gain loss of dynamic range, binning loss too
I am sure you cannot tell the difference with sum or average under 16bit mode

you can do one experiment to persuade me :D
just show me 2 images with the different method
you can increase gain by 120(4X=12db) to simulate sum up under 16bit
so you can take one image at 8bit at gain 200 and 16bit at gain 320
both at bin2 mode

_________________
ZWO Founder
Location:lon=120.6 lat=31.3
SuZhou China


Top 
 Profile  
Reply with quote  
 Post Posted: Wed Oct 11, 2017 7:46 pm 
Offline

Joined: Fri Nov 18, 2016 3:12 pm
Posts: 20
Sam,

I don't think you are reading my posts. I do use high gain. You are ignoring all the ADC issues I have described above. In my earlier posts I told you that I tried "higher" gain and it did not work. There were significant quantization and information losses.

Making me jump through hoops will not change the ADC issues that exist or my opinion or the fact that ZWO is using non standard operations and calling it binning.

Averaging just is not the best approach for deep sky imaging and I cannot think of one application where it is useful. Also there are no saturation issues with binning using Sum.

I am really puzzled as to why you are pushing back. The community is giving you clear feedback on the features they want. If you really want to keep the averaging feature then why not have both approaches and let the user decide. Why are you forcing your choice on the user and behaving in such a user unfriendly fashion?

If you want to fix this issue with your drivers then its fine otherwise I will take my business elsewhere. I am done arguing here.

-Hiten


Top 
 Profile  
Reply with quote  
 Post Posted: Thu Oct 12, 2017 7:45 am 
Offline
User avatar

Joined: Thu Feb 21, 2013 2:51 am
Posts: 2281
Hi Hiten
don't forget our company have many users too
and myself also user

we cannot find a reason to change it if you cannot convince me

Hs9 wrote:
I don't think you are reading my posts. I do use high gain. You are ignoring all the ADC issues I have described above. In my earlier posts I told you that I tried "higher" gain and it did not work. There were significant quantization and information losses.

of course I read your post and did replied back, "higher"gain means add 120(12db) to the original gain
this can do exactly the same effect as sum up under RAW16bit
I would like to see the image if you can test it and say there is difference

Hs9 wrote:
Making me jump through hoops will not change the ADC issues that exist or my opinion or the fact that ZWO is using non standard operations and calling it binning.

Averaging just is not the best approach for deep sky imaging and I cannot think of one application where it is useful. Also there are no saturation issues with binning using Sum.


I don't think there is ADC issue, for example
A 60
B 60
C 60
D 60
SUM UP 240

AFTER GAIN 4X
A 240
B 240
C 240
D 240
AVERAGE 240

please tell me the principle why sum is better Hiten
we are manufacturer and we need to know why

_________________
ZWO Founder
Location:lon=120.6 lat=31.3
SuZhou China


Top 
 Profile  
Reply with quote  
 Post Posted: Thu Oct 12, 2017 3:18 pm 
Offline

Joined: Fri Nov 18, 2016 3:12 pm
Posts: 20
I have already explained the reasons multiple times above. You keep ignoring my posts and throwing these simplistic examples at me. Once again... I am already operating at relatively high gain 350-400+ using short exposures 5-10s to minimize read noise. Going any higher causes all the ADC issues I have described above. Further, your simple example does not address amp-glow, posterization, loss of dynamic range / information loss and clipping issues that arise from using high gain.

Further, averaging is a non standard operation in Astro deepsky applications where we are operating in a photon starved environment. For example when doing Photometry on faint stars sum is incredibly useful in measuring flux. Further, most applications / file formats don't store float values which means I lose a lot of information / levels when capturing in bin x2 mode.

Another very important issue here is which I discussed above is... you should not make these decisions on the user's behalf. As a camera manufacturer you should stick with standard operations. Using averaging is non-standard and causes confusion. I have not seen one user explicitly ask for averaging. In fact everyone is constantly confused as to why binning for ZWO cameras is not working the way it should. Having brought numerous technology products to market in the last 3 decades I would say that this alone is reason enough to fix this.

How difficult is it for you to support both options? I would imagine this is a pretty easy change. I have 25+ years of technology and Semiconductor industry experience and have led numerous product development efforts so speak from experience. Which is why I am so surprised at your reluctance to listen to your users.

Sorry but I am really not understanding why you are pushing back. If it was a hard expensive change I would understand but this is not more than a few man hours. Give me the SDK code and I will do it for you.


Top 
 Profile  
Reply with quote  
 Post Posted: Sun Oct 15, 2017 4:10 pm 
Offline

Joined: Mon Aug 17, 2015 10:53 pm
Posts: 6
XS_MAN wrote:
I do planetary and lunar imaging. Never used binning for
this kind of target and I don't know any person using binning for this kind
of target.
Albéric


Hi All,
I used the binning often, sometimes desperately also, for reasons I won't tell, on planetary targets.
Besides, a suggestion about binning for planetary works can be found here:
http://www.planetary-astronomy-and-imag ... images.pdf

Interesting discussion, with remarkable and very instructive considerations.
I tried to read everything carefully.

I read a textline about the value of the opinions expressed by the users community, among the others. Now, if I can speak freely, I would like to express my preference in favour of a standard procedure.

Greetings


Last edited by laserjam on Wed Oct 18, 2017 7:55 pm, edited 2 times in total.

Top 
 Profile  
Reply with quote  
 Post Posted: Wed Oct 18, 2017 8:41 am 
Offline
User avatar

Joined: Thu Feb 21, 2013 2:51 am
Posts: 2281
Hi Hiten
I agree with you that it may cause misunderstand for Photometry
so I think the best way to bin under 16bit is sum up and don't shift 2 bits after that
but it lose dynamic range if we do like that
so which is better?

_________________
ZWO Founder
Location:lon=120.6 lat=31.3
SuZhou China


Top 
 Profile  
Reply with quote  
 Post Posted: Wed Oct 18, 2017 4:56 pm 
Offline

Joined: Fri Nov 18, 2016 3:12 pm
Posts: 20
Thanks Sam.

With regards to shifting bits and dynamic range, here is my opinion from 25 years of imaging (both pretty picture and scientific). I have respect for you and ZWO and I hope this does not come across as me preaching but this is a very important point to make.

The most important unwritten principle for most of us who use dedicated astro cameras is that the output should be untouched / unmodified / RAW to the largest extent possible. This and cooling are the two most important reasons we pick dedicated astro cameras over DSLRs and video cameras where the in built DSP wants to decide what is best for the user. Hence I prefer you don't shift the bits.

Also note that there will be no loss of dynamic range for a 12 bit ADC even in the extreme case assuming complete saturation of a single exposure even when using x4 bin.

A 16 bit image can fit 2^16 = 65,536 levels. So even x4 bin (sum) output with a 12 bit ADC will fit in a Mono16 format.

12 bit ADC using Mono 16 - Assumes extreme case where the frame has saturated pixels

Single exposure x1 bin = 2^12 = 4,096 levels
Single exposure x2 bin = 4,096 x 4 = 16,384 levels
Single exposure x3 bin = 4,096 x 9 = 36,864 levels
Single exposure x4 bin = 4,096 x 16 = 65,536 levels

There will be no loss of dynamic range even with x4 bin.

But ultimately the most important principle is that the output should be as RAW as possible. The decision of how to deal with saturation in extreme binning situations should be left to the user as there may be use cases which you cannot envision (Photometry is just a more obvious example).


Top 
 Profile  
Reply with quote  
 Post Posted: Thu Oct 19, 2017 10:12 am 
Offline
User avatar

Joined: Thu Feb 21, 2013 2:51 am
Posts: 2281
now we use the high 12bits of 16bits to save the image
because most software choose to display the image with the high 8bits
then user can see the same image under raw8 or raw16
then bin2 will saturate if you don't shift it to right

I think we'd better shift it by 2bits with bin2
and 3bits with bin3
4bits with bin4
but from customer's view, the brightness won't change like the average algorithm we use right now

what do you think?

_________________
ZWO Founder
Location:lon=120.6 lat=31.3
SuZhou China


Top 
 Profile  
Reply with quote  
 Post Posted: Thu Oct 19, 2017 3:07 pm 
Offline

Joined: Fri Nov 18, 2016 3:12 pm
Posts: 20
Sam,

My preference is for you not to shift the bits in the first place so that you don't have to shift them back. I would just take the sensor output, bin it and then use the full 16bits of dynamic range.

If the bit shifting happens in the driver rather than in the SDK I would first reverse the operation in the SDK i.e. shift the bits back and then bin and then use the full 16 bit output.

Binning can potentially double/triple/quadruple the dynamic range for x2/x3/x4 binning respectively. So if you shift the bits after binning you will lose dynamic range and most importantly information in the stack. A x4 bin 12bit image could have information in all the 16bits for example. So if you discard the higher bits you will lose information and cause clipping.

Most software that can display or process 16 bit images usually has the ability to stretch histograms - so no need to shift bits. For 8 bit this is a non issue.

Hiten


Top 
 Profile  
Reply with quote  
 Post Posted: Fri Oct 20, 2017 8:29 am 
Offline
User avatar

Joined: Thu Feb 21, 2013 2:51 am
Posts: 2281
Hs9 wrote:
Sam,
If the bit shifting happens in the driver rather than in the SDK I would first reverse the operation in the SDK i.e. shift the bits back and then bin and then use the full 16 bit output.
Hiten

actually, this is what we do currently and we call it average algorithm
take bin2 for example
shift the RAW data left 4 bits output as 16bit data
sum up 4 pixels and shift the bits back 2 bits
this operation just used the full 16bit output

_________________
ZWO Founder
Location:lon=120.6 lat=31.3
SuZhou China


Top 
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
 
Post new topic Reply to topic  [ 51 posts ]  Go to page Previous  1, 2, 3, 4, 5, 6  Next

Board index » Products » Cameras


Who is online

Users browsing this forum: Bing [Bot], Google [Bot], Yahoo [Bot] and 4 guests

 
 

 
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to: