Photometrics Prime95B Live vs Acquisition Speed

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Photometrics Prime95B Live vs Acquisition Speed

Chris O'Connell

Hi All,

 

I’m demoing a Photometrics Prime 95B camera.  It runs at the published frame rates when running Live.  When I capture a sequence it hits a bottleneck of about 115-120 fps no matter what size ROI I have.  This is true whether I write to our RAID data drive or RAM during capture. Here is what I get with 200 MHz, 12 bit readout, exposure set to 0 ms, and 5000 frames time lapse:

 

1200 x 1200: Live 80 fps – Acquisition 80 fps

 

1200 x 512: Live 190 fps – acquisition 115 fps

 

1200 x 256: Live 370 fps – Acquisition 117 fps

 

There are a lot of settings for this camera, so it’s not clear to me if the limiting factor is the computer (an older, but high end PC for our Andor spinning disk system) or some camera setting (buffer, etc.).  I tend to think it’s a camera setting since we run our Andor iXon 168 fps no problem.

If someone has any ideas, glad to hear them.

 

 

Chris O’Connell, PhD

Director, Advanced Light Microscopy Facility

 

University of Connecticut

Department of Physiology & Neurobiology

75 North Eagleville Road, U-3156

Storrs, CT 06269-3156

Office: 860-486-3271

[hidden email]

www.confocal.uconn.edu

 

 


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|

Re: Photometrics Prime95B Live vs Acquisition Speed

Nico Stuurman-2
Hi Chris,

On 1/31/18 7:46 AM, O'Connell, Christopher wrote:
> I’m demoing a Photometrics Prime 95B camera.  It runs at the published
> frame rates when running Live.  When I capture a sequence it hits a
> bottleneck of about 115-120 fps no matter what size ROI I have.

This may be obvious, but how do you measure sequence fps?


> This is true whether I write to our RAID data drive or RAM during
> capture. Here is what I get with 200 MHz, 12 bit readout, exposure set
> to 0 ms, and 5000 frames time lapse:
>
> 1200 x 1200: Live 80 fps – Acquisition 80 fps
>
> 1200 x 512: Live 190 fps – acquisition 115 fps
>
> 1200 x 256: Live 370 fps – Acquisition 117 fps
>
> There are a lot of settings for this camera, so it’s not clear to me
> if the limiting factor is the computer (an older, but high end PC for
> our Andor spinning disk system) or some camera setting (buffer,
> etc.).  I tend to think it’s a camera setting since we run our Andor
> iXon 168 fps no problem.
>

Best,

Nico

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|

Re: Photometrics Prime95B Live vs Acquisition Speed

Chris O'Connell
Hi Nico,

I'm looking at the fps displayed in MM.  We are using MM 1.4.23.  Interestingly, I just tried MM 2.0 beta and the live and acquisition fps speeds of the Prime 95B match and the data are now captured at the max rate.

We also have a similar issue with our Andor iXon DU897, where it would not reach published rates in MM 1.4.23, but will do it in 2.0 beta.  What's the difference?  Again, I'm basing all my measurements on the displayed fps.

Chris O'Connell, PhD
Director, Advanced Light Microscopy Facility

University of Connecticut
Department of Physiology & Neurobiology
75 North Eagleville Road, U-3156
Storrs, CT 06269-3156
Office: 860-486-3271
[hidden email]
www.confocal.uconn.edu



-----Original Message-----
From: Nico Stuurman [mailto:[hidden email]]
Sent: Wednesday, January 31, 2018 1:15 PM
To: Micro-Manager General <[hidden email]>
Subject: Re: [micro-manager-general] Photometrics Prime95B Live vs Acquisition Speed

Hi Chris,

On 1/31/18 7:46 AM, O'Connell, Christopher wrote:
> I'm demoing a Photometrics Prime 95B camera.  It runs at the published
> frame rates when running Live.  When I capture a sequence it hits a
> bottleneck of about 115-120 fps no matter what size ROI I have.

This may be obvious, but how do you measure sequence fps?


> This is true whether I write to our RAID data drive or RAM during
> capture. Here is what I get with 200 MHz, 12 bit readout, exposure set
> to 0 ms, and 5000 frames time lapse:
>
> 1200 x 1200: Live 80 fps - Acquisition 80 fps
>
> 1200 x 512: Live 190 fps - acquisition 115 fps
>
> 1200 x 256: Live 370 fps - Acquisition 117 fps
>
> There are a lot of settings for this camera, so it's not clear to me
> if the limiting factor is the computer (an older, but high end PC for
> our Andor spinning disk system) or some camera setting (buffer, etc.). 
> I tend to think it's a camera setting since we run our Andor iXon 168
> fps no problem.
>

Best,

Nico

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|

Re: Photometrics Prime95B Live vs Acquisition Speed

oftakofta
One way to measure "true" FPS is to look at the timestamps of the images. The timestamp is embedded in the ome-tif and it can be accessed by opening the file(s) in a HEX-editor, such as notepad++ with the HEX editor plugin. A quick and dirty way of doing FPS measurement is to look at the timestamp of the first and last images of a series and then divide the difference between them by the number of frames. You can easily find the timestamps by searching the HEX of your image stack for the ANSI String "TimeStampMsec". Start the search from the top to get the first one, then move the cursor to the bottom and do the same search from the bottom to get the last one. 

Here is the specification for the OME-TIFF file format, in case anyone is curious about it.


I am also currently setting up a system with a Prine95B 25 mm FOV, I'd be glad to compare notes on the subject :)

peace,

Jens Eriksson
Sellin Imaging Platform
Department of Medical Biochemistry and Microbiology
Uppsala University
 

On Wed, Jan 31, 2018 at 8:06 PM O'Connell, Christopher <[hidden email]> wrote:
Hi Nico,

I'm looking at the fps displayed in MM.  We are using MM 1.4.23.  Interestingly, I just tried MM 2.0 beta and the live and acquisition fps speeds of the Prime 95B match and the data are now captured at the max rate.

We also have a similar issue with our Andor iXon DU897, where it would not reach published rates in MM 1.4.23, but will do it in 2.0 beta.  What's the difference?  Again, I'm basing all my measurements on the displayed fps.

Chris O'Connell, PhD
Director, Advanced Light Microscopy Facility

University of Connecticut
Department of Physiology & Neurobiology
75 North Eagleville Road, U-3156
Storrs, CT 06269-3156
Office: <a href="tel:(860)%20486-3271" value="+18604863271" target="_blank">860-486-3271
[hidden email]
www.confocal.uconn.edu



-----Original Message-----
From: Nico Stuurman [mailto:[hidden email]]
Sent: Wednesday, January 31, 2018 1:15 PM
To: Micro-Manager General <[hidden email]>
Subject: Re: [micro-manager-general] Photometrics Prime95B Live vs Acquisition Speed

Hi Chris,

On 1/31/18 7:46 AM, O'Connell, Christopher wrote:
> I'm demoing a Photometrics Prime 95B camera.  It runs at the published
> frame rates when running Live.  When I capture a sequence it hits a
> bottleneck of about 115-120 fps no matter what size ROI I have.

This may be obvious, but how do you measure sequence fps?


> This is true whether I write to our RAID data drive or RAM during
> capture. Here is what I get with 200 MHz, 12 bit readout, exposure set
> to 0 ms, and 5000 frames time lapse:
>
> 1200 x 1200: Live 80 fps - Acquisition 80 fps
>
> 1200 x 512: Live 190 fps - acquisition 115 fps
>
> 1200 x 256: Live 370 fps - Acquisition 117 fps
>
> There are a lot of settings for this camera, so it's not clear to me
> if the limiting factor is the computer (an older, but high end PC for
> our Andor spinning disk system) or some camera setting (buffer, etc.). 
> I tend to think it's a camera setting since we run our Andor iXon 168
> fps no problem.
>

Best,

Nico

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|

Re: Photometrics Prime95B Live vs Acquisition Speed

Nico Stuurman-2
In reply to this post by Chris O'Connell
Hi Chris,

On 1/31/2018 10:34 AM, O'Connell, Christopher wrote:
> I'm looking at the fps displayed in MM.  We are using MM 1.4.23.  Interestingly, I just tried MM 2.0 beta and the live and acquisition fps speeds of the Prime 95B match and the data are now captured at the max rate.

I am still confused.  When you run an acquisition, MM does not display
the fps of the acquisition anywhere.  The various MM version do have fps
indicators signaling how fast the display updates (and for live mode,
how fast images were received).  Recording the time point of the first
and last frame is a better approach (as Jens suggested), but since that
is based on when MM received the image rather than when the exposure
happened, it is even better to look at hardware timestamps in the
metadata (also as Jens proposed), that are provided by some, but not
all, supported cameras.

In general, when the camera is running in sequence mode during an
acquisition, it will run at the speed allowed by the hardware.  When MM
drops frames, it will stop (and when the camera/driver drop frames, they
also should stop, but whether or not that is the case depends on the
camera driver/adapter code).  The fps indicators you are looking at, are
more likely indicators of how fast the images are displayed on the
screen (and some of the details changed between 1.4 and 2.0).

Best,

Nico



------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general