not (yet) supported cameras in micro-manager - general question

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

not (yet) supported cameras in micro-manager - general question

André Lampe (FU Berlin)
Hi everybody,

I started to take a closer look at the hobbyist side of microscopy,
especially affordable camera-solutions, since I am working more and more
in the science communication field and I wanted to share my fascination
of the small world. I recommend micro manager often to hobbyists, since
most cameras can be used using the openCVgrabber. Now, I found some
interesting cameras from "Toup Tek" (company name something chineese
beginning with an "H") using the software "ToupView". I will provide
links at the bottom of this post.

Nice cams, affordable - good thing for a hobbyist and a nice thing for
pros equipping e.g. cell-culture scopes with a little camera using the
ocular. BUT: can not run them in micro-manager. Could anyone offer
advice on how to use those non-standard non-pro cameras or a "HowTo make
Cams usable in mm"?

Spec-Sheet of Cams read: Capture/Control API: Native C/C++, C#,
Directshow, Twain, Labview.

TWAIN, openCVgrabber and ... well all camera options in MM produced an
error massage. On the other hand, the cam was running fine in its
software ToupView.

I did search the mailing-list - a hint for a suitable search-phrase for
this would also be appreciated ;-)

Links:

http://www.touptek.com/product/product.php?lang=en&class2=98

https://teleskop-austria.at/MicroqW320_32-Mpixel-MicroQ-W-widefield-PRO-digital-Mikroskop#m

Cheers, André

____________________________________________
Dr. André Lampe (Berlin)
Scientist - Presenter - Science Communicator
Twitter: @andereLampe
Blog (german): scienceblogs.de/diekleinendinge/

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Roy G. Biv
I think that the AmScope cameras are rebranded Touptek cameras. At least a few of the AmScope cameras are supposed to work together with micromanager. I first wanted to use the Toupview software (or AmScopes rebranded version) for horizontal line measurements but discovered that this wasn't possible in live view. I have not tried MM yet.

Dan



Den 9 dec. 2017 19:34 skrev "André Lampe (FU Berlin)" <[hidden email]>:
Hi everybody,

I started to take a closer look at the hobbyist side of microscopy, especially affordable camera-solutions, since I am working more and more in the science communication field and I wanted to share my fascination of the small world. I recommend micro manager often to hobbyists, since most cameras can be used using the openCVgrabber. Now, I found some interesting cameras from "Toup Tek" (company name something chineese beginning with an "H") using the software "ToupView". I will provide links at the bottom of this post.

Nice cams, affordable - good thing for a hobbyist and a nice thing for pros equipping e.g. cell-culture scopes with a little camera using the ocular. BUT: can not run them in micro-manager. Could anyone offer advice on how to use those non-standard non-pro cameras or a "HowTo make Cams usable in mm"?

Spec-Sheet of Cams read: Capture/Control API: Native C/C++, C#, Directshow, Twain, Labview.

TWAIN, openCVgrabber and ... well all camera options in MM produced an error massage. On the other hand, the cam was running fine in its software ToupView.

I did search the mailing-list - a hint for a suitable search-phrase for this would also be appreciated ;-)

Links:

http://www.touptek.com/product/product.php?lang=en&class2=98

https://teleskop-austria.at/MicroqW320_32-Mpixel-MicroQ-W-widefield-PRO-digital-Mikroskop#m

Cheers, André

____________________________________________
Dr. André Lampe (Berlin)
Scientist - Presenter - Science Communicator
Twitter: @andereLampe
Blog (german): scienceblogs.de/diekleinendinge/

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Geoff Pickford
Hi Dan and Andre,

I have been successful in linking the AmScope MU503B with MM, and I can now do a live Plot Profile on a live image  so as to find the very best focus point - as well as measure gray levels.  However, as I pointed out in my recent post, there is a problem with the embedded ImageJ Plot Profile as there is not a 1:1 correspondence between the cursor position and graph scale as soon as the ‘Y’ axis changes due to a change is focus etc.  To overcome this, I simply focus using Plot Profile, close the Plot Profile window and re-start it again to achieve 1:1 correspondence,  The cursor then matches the plot “y” values.

I would however like to find out how to fix this and also to link up a Tuscsen ISH500 camera to MM.  I will also try OpenCVgrabber.


Geoff Pickford


Pickford Resources Pty Ltd
PO Box 618, Wollongong, NSW, 2500
Laboratory Fax:  02 4227 6673.  Mobile:  0417 417 507

On 10 Dec 2017, at 6:44 AM, Roy G. Biv <[hidden email]> wrote:

I think that the AmScope cameras are rebranded Touptek cameras. At least a few of the AmScope cameras are supposed to work together with micromanager. I first wanted to use the Toupview software (or AmScopes rebranded version) for horizontal line measurements but discovered that this wasn't possible in live view. I have not tried MM yet.

Dan



Den 9 dec. 2017 19:34 skrev "André Lampe (FU Berlin)" <[hidden email]>:
Hi everybody,

I started to take a closer look at the hobbyist side of microscopy, especially affordable camera-solutions, since I am working more and more in the science communication field and I wanted to share my fascination of the small world. I recommend micro manager often to hobbyists, since most cameras can be used using the openCVgrabber. Now, I found some interesting cameras from "Toup Tek" (company name something chineese beginning with an "H") using the software "ToupView". I will provide links at the bottom of this post.

Nice cams, affordable - good thing for a hobbyist and a nice thing for pros equipping e.g. cell-culture scopes with a little camera using the ocular. BUT: can not run them in micro-manager. Could anyone offer advice on how to use those non-standard non-pro cameras or a "HowTo make Cams usable in mm"?

Spec-Sheet of Cams read: Capture/Control API: Native C/C++, C#, Directshow, Twain, Labview.

TWAIN, openCVgrabber and ... well all camera options in MM produced an error massage. On the other hand, the cam was running fine in its software ToupView.

I did search the mailing-list - a hint for a suitable search-phrase for this would also be appreciated ;-)

Links:

http://www.touptek.com/product/product.php?lang=en&class2=98

https://teleskop-austria.at/MicroqW320_32-Mpixel-MicroQ-W-widefield-PRO-digital-Mikroskop#m

Cheers, André

____________________________________________
Dr. André Lampe (Berlin)
Scientist - Presenter - Science Communicator
Twitter: @andereLampe
Blog (german): scienceblogs.de/diekleinendinge/

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

André Lampe (FU Berlin)
Hi Geoff, Dan and everybody,

I got the cam to run!

For the cam I got, which is a E3CMOS Series C-mount USB3.0 CMOS Camera E3CMOS02300KPA EP102300A, 2.3MegaPixel, Sony back illum. Chip it worked as fallows:
There are two general drivers for all Toup Tek Cams from Nov 21st 2016, one "DirectShow" and one TWAIN driver. Installing the TWAIN driver did not help, MM 1.4 and 2.0 beta3 crashed on me when adding the cam, but I think this is due to TWAIN not working (could not find the post again, talking about this). Anyway, installing the latest DirctShow driver lets you use the cam with openCVgrabber. only downside of this: I can not change "resolution" in the device property manager. The cam has two settings, 1920x1200 and 960x600 (for more FPS). The latter 960x600 is not an option in the drop down there, so the cam is always set to 1920x1200. With that setting I get the same fps as in ToupView, which is around 6-8 fps, way lower than the promised 38fps, sadly. But I will play around with it some more.

Links to drivers:
http://www.teleskop-austria.at/information/pdf/MicroQ_Driver-ToupTekToupcamDshowSetup.exe
http://www.teleskop-austria.at/information/pdf/MicroQ_Driver-ToupTekToupcamTwainSetup.exe

In any case: anyone tinkering around with hobbyist stuff feel free to contact me. I would like to hear more about using MM not only in a professional lab context.

Cheers, André

Am 10.12.2017 um 08:09 schrieb Geoff Pickford:
Hi Dan and Andre,

I have been successful in linking the AmScope MU503B with MM, and I can now do a live Plot Profile on a live image  so as to find the very best focus point - as well as measure gray levels.  However, as I pointed out in my recent post, there is a problem with the embedded ImageJ Plot Profile as there is not a 1:1 correspondence between the cursor position and graph scale as soon as the ‘Y’ axis changes due to a change is focus etc.  To overcome this, I simply focus using Plot Profile, close the Plot Profile window and re-start it again to achieve 1:1 correspondence,  The cursor then matches the plot “y” values.

I would however like to find out how to fix this and also to link up a Tuscsen ISH500 camera to MM.  I will also try OpenCVgrabber.


Geoff Pickford


Pickford Resources Pty Ltd
PO Box 618, Wollongong, NSW, 2500
Laboratory Fax:  02 4227 6673.  Mobile:  0417 417 507

On 10 Dec 2017, at 6:44 AM, Roy G. Biv <[hidden email]> wrote:

I think that the AmScope cameras are rebranded Touptek cameras. At least a few of the AmScope cameras are supposed to work together with micromanager. I first wanted to use the Toupview software (or AmScopes rebranded version) for horizontal line measurements but discovered that this wasn't possible in live view. I have not tried MM yet.

Dan



------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Karl Bellve-3


On Mon, Dec 11, 2017 at 11:28 AM André Lampe (FU Berlin) <[hidden email]> wrote:
Hi Geoff, Dan and everybody,

I got the cam to run!

For the cam I got, which is a E3CMOS Series C-mount USB3.0 CMOS Camera E3CMOS02300KPA EP102300A, 2.3MegaPixel, Sony back illum. Chip it worked as fallows:
There are two general drivers for all Toup Tek Cams from Nov 21st 2016, one "DirectShow" and one TWAIN driver. Installing the TWAIN driver did not help, MM 1.4 and 2.0 beta3 crashed on me when adding the cam, but I think this is due to TWAIN not working (could not find the post again, talking about this). Anyway, installing the latest DirctShow driver lets you use the cam with openCVgrabber. only downside of this: I can not change "resolution" in the device property manager. The cam has two settings, 1920x1200 and 960x600 (for more FPS). The latter 960x600 is not an option in the drop down there, so the cam is always set to 1920x1200. With that setting I get the same fps as in ToupView, which is around 6-8 fps, way lower than the promised 38fps, sadly. But I will play around with it some more.

Links to drivers:
http://www.teleskop-austria.at/information/pdf/MicroQ_Driver-ToupTekToupcamDshowSetup.exe
http://www.teleskop-austria.at/information/pdf/MicroQ_Driver-ToupTekToupcamTwainSetup.exe

In any case: anyone tinkering around with hobbyist stuff feel free to contact me. I would like to hear more about using MM not only in a professional lab context.

Cheers, André

Am 10.12.2017 um 08:09 schrieb Geoff Pickford:
Hi Dan and Andre,

I have been successful in linking the AmScope MU503B with MM, and I can now do a live Plot Profile on a live image  so as to find the very best focus point - as well as measure gray levels.  However, as I pointed out in my recent post, there is a problem with the embedded ImageJ Plot Profile as there is not a 1:1 correspondence between the cursor position and graph scale as soon as the ‘Y’ axis changes due to a change is focus etc.  To overcome this, I simply focus using Plot Profile, close the Plot Profile window and re-start it again to achieve 1:1 correspondence,  The cursor then matches the plot “y” values.

I would however like to find out how to fix this and also to link up a Tuscsen ISH500 camera to MM.  I will also try OpenCVgrabber.


Geoff Pickford


Pickford Resources Pty Ltd
PO Box 618, Wollongong, NSW, 2500
Laboratory Fax:  02 4227 6673.  Mobile:  0417 417 507

On 10 Dec 2017, at 6:44 AM, Roy G. Biv <[hidden email]> wrote:

I think that the AmScope cameras are rebranded Touptek cameras. At least a few of the AmScope cameras are supposed to work together with micromanager. I first wanted to use the Toupview software (or AmScopes rebranded version) for horizontal line measurements but discovered that this wasn't possible in live view. I have not tried MM yet.

Dan



Hi Dan, André, Geoff and everyone else,

I am following this conversation closely. What would you recommend for <$500 CCD camera? Monochrome or Color but not CMOS. I think non scientific CMOS cameras are there just yet. Mainly for studying GFP in worms for a lab that I don't think needs to spend thousands to do this.



Cheers Karl 

Karl Bellvé
Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical School

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Nico Stuurman-2
Hi Karl,

On 12/12/2017 8:09 AM, Karl Bellve wrote:
I am following this conversation closely. What would you recommend for <$500 CCD camera? Monochrome or Color but not CMOS. I think non scientific CMOS cameras are there just yet. Mainly for studying GFP in worms for a lab that I don't think needs to spend thousands to do this.

What is the problem with non-scientific CMOS?  We have a couple of the POint grey (now FLIR) CMOS cameras, and they provide perform very well.  Our standard epi scope now has a Chameleon3 (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2 ER (which at the time had the lowest read-noise possible at 6 electrons) and nobody complains (actually, everyone loves the speed).  Only downside is that these do not have cooling and run quite hot, and hot pixels will be visible on long exposures (but when do you ever need second long exposures?).  If I remember correctly we paid something like $400 for this.

The CCD in this same series (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is more expensive, has 10.5 electrons read noise, and runs at no more than 13 fps. 

There are now a couple of papers describing single molecule imaging with non-scientific CMOS cameras (https://www.biorxiv.org/content/early/2017/09/09/186544, https://www.nature.com/articles/s41598-017-14762-6).

Curious to hear what you do not like about them.

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: not (yet) supported cameras in micro-manager - general question

Karl Bellve-3


On Tue, Dec 12, 2017 at 11:47 AM Nico Stuurman <[hidden email]> wrote:
Hi Karl,


On 12/12/2017 8:09 AM, Karl Bellve wrote:
I am following this conversation closely. What would you recommend for <$500 CCD camera? Monochrome or Color but not CMOS. I think non scientific CMOS cameras are there just yet. Mainly for studying GFP in worms for a lab that I don't think needs to spend thousands to do this.

What is the problem with non-scientific CMOS?  We have a couple of the POint grey (now FLIR) CMOS cameras, and they provide perform very well.  Our standard epi scope now has a Chameleon3 (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2 ER (which at the time had the lowest read-noise possible at 6 electrons) and nobody complains (actually, everyone loves the speed).  Only downside is that these do not have cooling and run quite hot, and hot pixels will be visible on long exposures (but when do you ever need second long exposures?).  If I remember correctly we paid something like $400 for this.

The CCD in this same series (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is more expensive, has 10.5 electrons read noise, and runs at no more than 13 fps. 

There are now a couple of papers describing single molecule imaging with non-scientific CMOS cameras (https://www.biorxiv.org/content/early/2017/09/09/186544, https://www.nature.com/articles/s41598-017-14762-6).

Curious to hear what you do not like about them.

Best,

Nico


Nico and André

We just had a discussion on non scientific CMOS and CCD cameras yesterday in our small microscopy group.

The nature paper you linked was brought up.

Look at the histogram of gain in Figure 1i in the Nature paper. Also look at the noise in 1e. Both of these could be very difficult to quantify and calibrate for the average microscopist. Also, look at Figure 2c. The variance is in localization is massive for the CMOS vs sCMOS in low to mid photon applications. 

My take home message is non-scientific CMOS cameras are not linear, and maybe difficult to calibrate for quantitative imaging. Industrial grade CCD cameras don't appear to have the same issues.

So, I was thinking that for the average microscopist, I would recommend industrial CCD over industrial CMOS. But, I am new to these cameras...and still learning. :D 



Cheers Karl

Karl Bellvé

Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical School

 

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Nico Stuurman-2

On 12/12/17 9:10 AM, Karl Bellve wrote:

>
> On Tue, Dec 12, 2017 at 11:47 AM Nico Stuurman <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>
>     On 12/12/2017 8:09 AM, Karl Bellve wrote:
>>     I am following this conversation closely. What would you
>>     recommend for <$500 CCD camera? Monochrome or Color but not CMOS.
>>     I think non scientific CMOS cameras are there just yet. Mainly
>>     for studying GFP in worms for a lab that I don't think needs to
>>     spend thousands to do this.
>
>     What is the problem with non-scientific CMOS?  We have a couple of
>     the POint grey (now FLIR) CMOS cameras, and they provide perform
>     very well.  Our standard epi scope now has a Chameleon3
>     (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and
>     runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2
>     ER (which at the time had the lowest read-noise possible at 6
>     electrons) and nobody complains (actually, everyone loves the
>     speed).  Only downside is that these do not have cooling and run
>     quite hot, and hot pixels will be visible on long exposures (but
>     when do you ever need second long exposures?).  If I remember
>     correctly we paid something like $400 for this.
>
>     The CCD in this same series
>     (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is
>     more expensive, has 10.5 electrons read noise, and runs at no more
>     than 13 fps.
>
>     There are now a couple of papers describing single molecule
>     imaging with non-scientific CMOS cameras
>     (https://www.biorxiv.org/content/early/2017/09/09/186544,
>     https://www.nature.com/articles/s41598-017-14762-6).
>
>
>
> We just had a discussion on non scientific CMOS and CCD cameras
> yesterday in our small microscopy group.
>
> The nature paper you linked was brought up.
>
> Look at the histogram of gain in Figure 1i in the Nature paper. Also
> look at the noise in 1e. Both of these could be very difficult to
> quantify and calibrate for the average microscopist. Also, look at
> Figure 2c. The variance is in localization is massive for the CMOS vs
> sCMOS in low to mid photon applications.
>
> My take home message is non-scientific CMOS cameras are not linear,
> and maybe difficult to calibrate for quantitative imaging. Industrial
> grade CCD cameras don't appear to have the same issues.
>
> So, I was thinking that for the average microscopist, I would
> recommend industrial CCD over industrial CMOS. But, I am new to these
> cameras...and still learning. :D

The pixel gain variance in Fig. 1idoes look funny, but the spread is
about 10% (i.e. not that difference from the sCMOS).  For the average
biologist, who is not accurate flat-fielding the image in any case, and
just wants to see GFP fluorescence, that does not look like a big
problem (especially considering the price tag and speed of that camera).


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: not (yet) supported cameras in micro-manager - general question

Knecht, David
In reply to this post by Karl Bellve-3
I would agree with Karl on CCD vs. CMOS and to some extent sCMOS due to the inability of CMOS cameras to do binning the way CCD cameras do.  We can stretch the sensitivity of our cheap industrial CCD cameras by binning for weak fluorescence.  You don’t gain (excuse the pun) much if anything in a CMOS by binning.  We rarely have to go fast enough to have the speed of CMOS become valuable.  Dave

Dr. David Knecht
Professor , Department of Molecular and Cell Biology 
University of Connecticut
91 N. Eagleville Rd.
U-3125
Storrs, CT 06269-3125


On Dec 12, 2017, at 12:10 PM, Karl Bellve <[hidden email]> wrote:



On Tue, Dec 12, 2017 at 11:47 AM Nico Stuurman <[hidden email]> wrote:
Hi Karl,


On 12/12/2017 8:09 AM, Karl Bellve wrote:
I am following this conversation closely. What would you recommend for <$500 CCD camera? Monochrome or Color but not CMOS. I think non scientific CMOS cameras are there just yet. Mainly for studying GFP in worms for a lab that I don't think needs to spend thousands to do this.

What is the problem with non-scientific CMOS?  We have a couple of the POint grey (now FLIR) CMOS cameras, and they provide perform very well.  Our standard epi scope now has a Chameleon3 (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2 ER (which at the time had the lowest read-noise possible at 6 electrons) and nobody complains (actually, everyone loves the speed).  Only downside is that these do not have cooling and run quite hot, and hot pixels will be visible on long exposures (but when do you ever need second long exposures?).  If I remember correctly we paid something like $400 for this.

The CCD in this same series (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is more expensive, has 10.5 electrons read noise, and runs at no more than 13 fps.  

There are now a couple of papers describing single molecule imaging with non-scientific CMOS cameras (https://www.biorxiv.org/content/early/2017/09/09/186544, https://www.nature.com/articles/s41598-017-14762-6).

Curious to hear what you do not like about them.

Best,

Nico


Nico and André

We just had a discussion on non scientific CMOS and CCD cameras yesterday in our small microscopy group.

The nature paper you linked was brought up.

Look at the histogram of gain in Figure 1i in the Nature paper. Also look at the noise in 1e. Both of these could be very difficult to quantify and calibrate for the average microscopist. Also, look at Figure 2c. The variance is in localization is massive for the CMOS vs sCMOS in low to mid photon applications. 

My take home message is non-scientific CMOS cameras are not linear, and maybe difficult to calibrate for quantitative imaging. Industrial grade CCD cameras don't appear to have the same issues.

So, I was thinking that for the average microscopist, I would recommend industrial CCD over industrial CMOS. But, I am new to these cameras...and still learning. :D 



Cheers Karl

Karl Bellvé

Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical School

 
------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Karl Bellve-3
In reply to this post by Nico Stuurman-2


On Tue, Dec 12, 2017 at 1:25 PM Nico Stuurman <[hidden email]> wrote:

On 12/12/17 9:10 AM, Karl Bellve wrote:
>
> On Tue, Dec 12, 2017 at 11:47 AM Nico Stuurman <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>
>     On 12/12/2017 8:09 AM, Karl Bellve wrote:
>>     I am following this conversation closely. What would you
>>     recommend for <$500 CCD camera? Monochrome or Color but not CMOS.
>>     I think non scientific CMOS cameras are there just yet. Mainly
>>     for studying GFP in worms for a lab that I don't think needs to
>>     spend thousands to do this.
>
>     What is the problem with non-scientific CMOS?  We have a couple of
>     the POint grey (now FLIR) CMOS cameras, and they provide perform
>     very well.  Our standard epi scope now has a Chameleon3
>     (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and
>     runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2
>     ER (which at the time had the lowest read-noise possible at 6
>     electrons) and nobody complains (actually, everyone loves the
>     speed).  Only downside is that these do not have cooling and run
>     quite hot, and hot pixels will be visible on long exposures (but
>     when do you ever need second long exposures?).  If I remember
>     correctly we paid something like $400 for this.
>
>     The CCD in this same series
>     (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is
>     more expensive, has 10.5 electrons read noise, and runs at no more
>     than 13 fps.
>
>     There are now a couple of papers describing single molecule
>     imaging with non-scientific CMOS cameras
>     (https://www.biorxiv.org/content/early/2017/09/09/186544,
>     https://www.nature.com/articles/s41598-017-14762-6).
>
>
>
> We just had a discussion on non scientific CMOS and CCD cameras
> yesterday in our small microscopy group.
>
> The nature paper you linked was brought up.
>
> Look at the histogram of gain in Figure 1i in the Nature paper. Also
> look at the noise in 1e. Both of these could be very difficult to
> quantify and calibrate for the average microscopist. Also, look at
> Figure 2c. The variance is in localization is massive for the CMOS vs
> sCMOS in low to mid photon applications.
>
> My take home message is non-scientific CMOS cameras are not linear,
> and maybe difficult to calibrate for quantitative imaging. Industrial
> grade CCD cameras don't appear to have the same issues.
>
> So, I was thinking that for the average microscopist, I would
> recommend industrial CCD over industrial CMOS. But, I am new to these
> cameras...and still learning. :D

The pixel gain variance in Fig. 1idoes look funny, but the spread is
about 10% (i.e. not that difference from the sCMOS).  For the average
biologist, who is not accurate flat-fielding the image in any case, and
just wants to see GFP fluorescence, that does not look like a big
problem (especially considering the price tag and speed of that camera).



It is gain, a 10% factor could result in much larger apparent signal at higher photon counts. Since it is bimodal, are those gains spatially equally distributed or are they unequally distributed? It is possible that you could have a "hot" area of the camera and a "cool" area of the camera? sCMOS chips are usually hand selected to reduce these issues, and then the gains are mapped, and normalized, at least that is what I think they do. 

In any case, I assume you like that camera and that is the camera you would pick for <$500? 

Its specifications, QE, dark/read noise, etc look really good. 

Cheers Karl

Karl Bellvé
Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical.


 

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Kyle Douglass

Hi Karl and Nico,

Great discussion so far!


Since it is bimodal, are those gains spatially equally distributed or are they unequally distributed?


Figure 1h shows the corresponding gain map, and they are definitely not equally distributed. It appears that there is a very strong fixed pattern noise, and it's worse than anything I've ever seen in the three different sCMOS cameras I've flat-fielded.


My take home message is non-scientific CMOS cameras are not linear, and maybe difficult to calibrate for quantitative imaging.


The gain histograms don't demonstrate the linearity of the camera, do they? To do this, I thought that you need to plot the pixel temporal variance vs. mean photon fluence on the pixels, and I don't see this plotted anywhere in the main figures. In my experience, these curves in even scientific CMOS cameras will begin to saturate near the point where the camera's gain scheduling causes the amplifier to switch from the low noise one to the one with high well capacity, which is around 2000 ADU on ours. This is especially annoying in STORM with organic dyes because it's relatively easy to hit a few pixels with enough photons that the pixels in the central part of the PSF is using one amplifier, whereas the pixels on the edge of the PSF are using the other. (I might be wrong about this, but one manufacturer informed me that the amplifier switching occurs on a per pixel basis, not on the whole chip.)


Maybe an advantage of regular CMOS over sCMOS is that there is only one amplifier for every pixel?


Just my 2 cents,

Kyle



Dr. Kyle M. Douglass
Post-doctoral Researcher
EPFL - The Laboratory of Experimental Biophysics
http://leb.epfl.ch/
http://kmdouglass.github.io

From: Karl Bellve <[hidden email]>
Sent: Tuesday, December 12, 2017 8:20 PM
To: Micro-Manager General
Subject: Re: [micro-manager-general] not (yet) supported cameras in micro-manager - general question
 


On Tue, Dec 12, 2017 at 1:25 PM Nico Stuurman <[hidden email]> wrote:

On 12/12/17 9:10 AM, Karl Bellve wrote:
>
> On Tue, Dec 12, 2017 at 11:47 AM Nico Stuurman <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>
>     On 12/12/2017 8:09 AM, Karl Bellve wrote:
>>     I am following this conversation closely. What would you
>>     recommend for <$500 CCD camera? Monochrome or Color but not CMOS.
>>     I think non scientific CMOS cameras are there just yet. Mainly
>>     for studying GFP in worms for a lab that I don't think needs to
>>     spend thousands to do this.
>
>     What is the problem with non-scientific CMOS?  We have a couple of
>     the POint grey (now FLIR) CMOS cameras, and they provide perform
>     very well.  Our standard epi scope now has a Chameleon3
>     (CM3-U3-31S4M-CS), which has 2048x1536 pixels, QE of 71%,  and
>     runs 55fps with 2.9 electrons read-noise.  It replaces an Orca 2
>     ER (which at the time had the lowest read-noise possible at 6
>     electrons) and nobody complains (actually, everyone loves the
>     speed).  Only downside is that these do not have cooling and run
>     quite hot, and hot pixels will be visible on long exposures (but
>     when do you ever need second long exposures?).  If I remember
>     correctly we paid something like $400 for this.
>
>     The CCD in this same series
>     (https://www.ptgrey.com/chameleon3-28-mp-mono-usb3-vision-2) is
>     more expensive, has 10.5 electrons read noise, and runs at no more
>     than 13 fps.
>
>     There are now a couple of papers describing single molecule
>     imaging with non-scientific CMOS cameras
>     (https://www.biorxiv.org/content/early/2017/09/09/186544,
>     https://www.nature.com/articles/s41598-017-14762-6).
>
>
>
> We just had a discussion on non scientific CMOS and CCD cameras
> yesterday in our small microscopy group.
>
> The nature paper you linked was brought up.
>
> Look at the histogram of gain in Figure 1i in the Nature paper. Also
> look at the noise in 1e. Both of these could be very difficult to
> quantify and calibrate for the average microscopist. Also, look at
> Figure 2c. The variance is in localization is massive for the CMOS vs
> sCMOS in low to mid photon applications.
>
> My take home message is non-scientific CMOS cameras are not linear,
> and maybe difficult to calibrate for quantitative imaging. Industrial
> grade CCD cameras don't appear to have the same issues.
>
> So, I was thinking that for the average microscopist, I would
> recommend industrial CCD over industrial CMOS. But, I am new to these
> cameras...and still learning. :D

The pixel gain variance in Fig. 1idoes look funny, but the spread is
about 10% (i.e. not that difference from the sCMOS).  For the average
biologist, who is not accurate flat-fielding the image in any case, and
just wants to see GFP fluorescence, that does not look like a big
problem (especially considering the price tag and speed of that camera).



It is gain, a 10% factor could result in much larger apparent signal at higher photon counts. Since it is bimodal, are those gains spatially equally distributed or are they unequally distributed? It is possible that you could have a "hot" area of the camera and a "cool" area of the camera? sCMOS chips are usually hand selected to reduce these issues, and then the gains are mapped, and normalized, at least that is what I think they do. 

In any case, I assume you like that camera and that is the camera you would pick for <$500? 

Its specifications, QE, dark/read noise, etc look really good. 

Cheers Karl

Karl Bellvé
Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical.


 

------------------------------------------------------------------------------
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: not (yet) supported cameras in micro-manager - general question

Karl Bellve-3


On Tue, Dec 12, 2017 at 3:08 PM Kyle Michael Douglass <[hidden email]> wrote:

Hi Karl and Nico,

Great discussion so far!


Since it is bimodal, are those gains spatially equally distributed or are they unequally distributed?


Figure 1h shows the corresponding gain map, and they are definitely not equally distributed. It appears that there is a very strong fixed pattern noise, and it's worse than anything I've ever seen in the three different sCMOS cameras I've flat-fielded.


My take home message is non-scientific CMOS cameras are not linear, and maybe difficult to calibrate for quantitative imaging.


The gain histograms don't demonstrate the linearity of the camera, do they? To do this, I thought that you need to plot the pixel temporal variance vs. mean photon fluence on the pixels, and I don't see this plotted anywhere in the main figures.


Maybe I wasn't using the right term, but I was more thinking the influence of having a large bimodal gain distribution and how individual pixels react to light compared to neighboring pixels in the same field. I would think it would more problematic to do quantitative imaging within an image (pixel to pixel comparison), rather than between images (pixel at same location over time). The gains have to be mapped. 

In my experience, these curves in even scientific CMOS cameras will begin to saturate near the point where the camera's gain scheduling causes the amplifier to switch from the low noise one to the one with high well capacity, which is around 2000 ADU on ours. This is especially annoying in STORM with organic dyes because it's relatively easy to hit a few pixels with enough photons that the pixels in the central part of the PSF is using one amplifier, whereas the pixels on the edge of the PSF are using the other. (I might be wrong about this, but one manufacturer informed me that the amplifier switching occurs on a per pixel basis, not on the whole chip.)


Maybe an advantage of regular CMOS over sCMOS is that there is only one amplifier for every pixel?


Good point.


Just my 2 cents,

Kyle




Cheers Karl

Karl Bellvé
Biomedical Imaging Group
Molecular Medicine
University of Massachusetts Medical School

------------------------------------------------------------------------------
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