Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Canon EOS RP sensor analysis

Astrophotography DSLR Equipment
  • Please log in to reply
29 replies to this topic

#1 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 14 April 2021 - 03:16 PM

I've ran the Sharpcap's  sensor analysis on my Canon EOS Rp (modified). It costed me 500+ shutter actuations. I know some people here on CN use this camera for astrophotography. This analysis should help deciding which ISO value is best for your particular needs. I've attached the resulting table and made a plot of the read noise and the full well values.

As you can see, indeed the noise is very high at low ISO. Only at and after ISO 640 the noise becomes manageable. The full well values look good comparing to recent dedicated astronomy cameras. The dynamic range is so-so (around 11).

In my opinion for the EOS RP camera the ISO 1600 is a good compromise shooting DSO objects. 640 may also be good.
Oh, and Sharpcap is a really nice piece of software! I also analyzed my ZWO ASI183mc  and the values closely match the values from the ZWO website.

 

EOS Rp plot.jpg

EOS Rp chart.jpg


  • 2ghouls likes this

#2 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 14 April 2021 - 05:29 PM

Some of those read noise values are quite a lot higher than on PhotonsToPhotos, especially at high ISOs:

https://www.photonst...Canon EOS RP_14

 

Mark


Edited by sharkmelley, 14 April 2021 - 05:31 PM.

  • SandyHouTex likes this

#3 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,768
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 14 April 2021 - 06:37 PM

Some of those read noise values are quite a lot higher than on PhotonsToPhotos, especially at high ISOs:

https://www.photonst...Canon EOS RP_14

 

Mark

You beat me to it.  The highest I see at low ISOs (100) is about 5+ electrons.



#4 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 14 April 2021 - 06:44 PM

Some of those read noise values are quite a lot higher than on PhotonsToPhotos, especially at high ISOs:

https://www.photonst...Canon EOS RP_14

 

Mark

Mark, those are derived values, so I'm not sure how to compare them with mine.

Also, the difference is not large, ~2.5 vs 3.7. From that page ""Hi" ISO values can behave strangely once signal processing such as noise reduction kicks in."

Overall, values are close in the whole ISO range. 

 

Sergey.


Edited by astroserge, 14 April 2021 - 08:37 PM.


#5 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 15 April 2021 - 01:25 AM

Mark, those are derived values, so I'm not sure how to compare them with mine.

Also, the difference is not large, ~2.5 vs 3.7. From that page ""Hi" ISO values can behave strangely once signal processing such as noise reduction kicks in."

Overall, values are close in the whole ISO range. 

 

Sergey.

What do you mean by "derived values"?  Are you saying your values are not derived?

 

The difference at ISO 10000 is 3.7 vs 2.1  That's a pretty big difference in my book i.e. 76% higher.

A noise reading as high as 3.7e- RMS at high ISO is not impressive.

 

The differences could be due to Sharpcap's methodology - I don't know how its calculations work.  On the other hand maybe you have a genuinely noisy copy of the EOS RP sensor.

 

Mark


Edited by sharkmelley, 15 April 2021 - 01:36 AM.

  • SandyHouTex likes this

#6 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 15 April 2021 - 01:36 PM

What do you mean by "derived values"?  Are you saying your values are not derived?

 

The difference at ISO 10000 is 3.7 vs 2.1  That's a pretty big difference in my book i.e. 76% higher.

A noise reading as high as 3.7e- RMS at high ISO is not impressive.

 

The differences could be due to Sharpcap's methodology - I don't know how its calculations work.  On the other hand maybe you have a genuinely noisy copy of the EOS RP sensor.

 

Mark

I'm looking at the whole picture. At the PhotonstoPhotos page the noise at low ISO (if you interpolate) is pretty close. And I don't care about ISO 10000 because the full well there is really small thus I'm not going to use it. So in my view my and PTP data are close. And both agree that noise at 100 ISO is well above 5. The issue of unusually high noise at low ISO was noted by many though they didn't quantize it.

As of Sharpcap's methodology the values from my ASI183MC test match those from the ZWO webpage. So I believe I can trust it.


Edited by astroserge, 15 April 2021 - 03:01 PM.


#7 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 15 April 2021 - 03:19 PM

You beat me to it.  The highest I see at low ISOs (100) is about 5+ electrons.

I think you are misreading the graph. The vertical axis is a log scale which means the read noise is not slightly more than 5 but slightly more than 2 to the power of 5 i.e. more than 32 electrons.  This is more than twice as high as Canon's first entry level camera, the Canon Digital Rebel.  It's shockingly high for a modern camera.

 

Mark



#8 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 15 April 2021 - 04:17 PM

I think you are misreading the graph. The vertical axis is a log scale which means the read noise is not slightly more than 5 but slightly more than 2 to the power of 5 i.e. more than 32 electrons.  This is more than twice as high as Canon's first entry level camera, the Canon Digital Rebel.  It's shockingly high for a modern camera.

 

Mark

Oh, I didn't realize that SandyHouTex was referring to the PTP plot not to his/her own measurements.
If you hover your mouse over the dots on the plot you can read normal values.
I have an EOS 450D camera, now I'm thinking to do the tests for it and compare.


  • 2ghouls likes this

#9 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,768
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 15 April 2021 - 07:53 PM

Oh, I didn't realize that SandyHouTex was referring to the PTP plot not to his/her own measurements.
If you hover your mouse over the dots on the plot you can read normal values.
I have an EOS 450D camera, now I'm thinking to do the tests for it and compare.

Unfortunately you can’t hover an iPad.

 

Next time I go there I’ll use a Windows machine.



#10 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,768
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 15 April 2021 - 10:29 PM

So I went to DXOMARK, and they gave the RPs sensor high marks for low ISO performance.

 

I’m confused.



#11 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 16 April 2021 - 01:12 AM

So I went to DXOMARK, and they gave the RPs sensor high marks for low ISO performance.

I don't see a low ISO performance measure.  Are you referring to the Portrait, Landscape or Sports score, or something else?

 

This is what it says about the weak performance at low ISO ( https://www.dxomark....p-sensor-review ):

"Although the RP performs well overall, its weakest result is in its dynamic range response, at just under 12 EV at base ISO; the other full-frame mirrorless models comfortably surpass it. In some cases, the best sensors are as much as +2 EV or more ahead at base ISO."

 

Mark


Edited by sharkmelley, 16 April 2021 - 01:14 AM.


#12 kreegan99

kreegan99

    Explorer 1

  • -----
  • Posts: 83
  • Joined: 20 May 2009
  • Loc: Endwell, NY

Posted 16 April 2021 - 08:41 AM

So I'm trying to learn from this thread.  Thought I would make sure I'm reading this ok.

 

Comparing the Rp, Ra and 6D

 

 

 

Screenshot 2021-04-16 091416.png

 

- The Ra seems to be much better than the 6D and much, much better than the Rp at lower ISO's

- The sweet spot for all of these cameras are the mid ISO ranges for DSO astrophotography (roughly 800 - 4000)

- The Ra and 6D, really appear to be better cameras for low light conditions with respect to noise.

- So signal overwhelms noise, do these read noise readings really impact image quality in day-time photography when the lower ISO would be used?

 

Again, just making sure I'm reading this information correctly, and trying to get a better understanding.

 

And now a real learning question.  I have a ZWO 1600MM Pro.  I typically use it at 100% gain (default), which is something I'm starting to research just today when I accidentally found this thread.  I'm assuming 100% gain is more like high ISO (or maybe I have that backwards)?  And is it also true that there is a sweet spot on that camera, and should I back off that some to find a similar sweet spot?  So if I have a bright star in the shot, likely want to back off the gain, and shoot longer for more details?

 

Thank You!



#13 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,768
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 16 April 2021 - 09:19 AM

I don't see a low ISO performance measure.  Are you referring to the Portrait, Landscape or Sports score, or something else?

 

This is what it says about the weak performance at low ISO ( https://www.dxomark....p-sensor-review ):

"Although the RP performs well overall, its weakest result is in its dynamic range response, at just under 12 EV at base ISO; the other full-frame mirrorless models comfortably surpass it. In some cases, the best sensors are as much as +2 EV or more ahead at base ISO."

 

Mark

So it’s under the “Overall Performance” section, paragraph 3:

 

“The RP also performs well in our Sports (low-light ISO) score, achieving a very respectable ISO 2997, which puts it among the best-performing sensors in this particular category.”



#14 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 16 April 2021 - 10:56 AM

So it’s under the “Overall Performance” section, paragraph 3:

 

“The RP also performs well in our Sports (low-light ISO) score, achieving a very respectable ISO 2997, which puts it among the best-performing sensors in this particular category.”

That's the low-light score not the low ISO score!  It's measuring the performance in low light (where the RP is good) and not the performance of the low ISO (where the RP is not good).

 

Mark


Edited by sharkmelley, 16 April 2021 - 01:36 PM.

  • astroserge likes this

#15 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 16 April 2021 - 12:41 PM

I tested my EOS 450D. See attached image.

My conclusions. Yes, at low ISO the read noise of the RP sensor is much higher than the one from a 10 years old camera and overall very high. At 1600 (recommended for EOS 450D in the APT chart https://www.astropho...phy.app/EOS.php ) is still higher for the RP (5.5 vs 3.0). But the full well and the dynamic range are significantly better for the EOS RP. Also, the QE of the RP sensor (68%!) is two times higher than the one for the 450D. It also weighs less and has a modern circuitry. With the EOS RP you can do live view in APT at ISO 102400 which really helps. So I still take the RP over older cameras for the astrophotography (if money is not a big concern). And for the AP I recommend ISO 1600 for the EOS RP sensor. 

I used my EOS RP camera for daytime shoots and it looks good to me. The auto-focusing is fast and robust with the kit zoom lens. Thus it is a quite useable all-around camera.

Sergey.

 

EOS 450D.jpg



#16 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 16 April 2021 - 12:59 PM

 

 

 

- The Ra and 6D, really appear to be better cameras for low light conditions with respect to noise.

- So signal overwhelms noise, do these read noise readings really impact image quality in day-time photography when the lower ISO would be used?

Check my last post on the day-time performance. Also, there are many reviews of the RP which say that the camera is good overall.



#17 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 16 April 2021 - 01:07 PM

 

And now a real learning question.  I have a ZWO 1600MM Pro.  I typically use it at 100% gain (default), which is something I'm starting to research just today when I accidentally found this thread.  I'm assuming 100% gain is more like high ISO (or maybe I have that backwards)?  And is it also true that there is a sweet spot on that camera, and should I back off that some to find a similar sweet spot?  So if I have a bright star in the shot, likely want to back off the gain, and shoot longer for more details?

 

Thank You!

My understanding is (and I'm not an expert!) that you should never use a camera at maximum gain unless for framing and such. At max gain you have a very small full well and DR. Yes, there is a compromise usually somewhere in the middle. Many people use the "unity" gain (where 1e corresponds to 1 ADU, check #2 column in my tables). But overall just look at the noise plot and use gain value where the plot becomes more or less flat or changes from steep to slow.



#18 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 16 April 2021 - 01:22 PM

Also, the QE of the RP sensor (68%!) is two times higher than the one for the 450D.

Where did you find the QE figure of 68%?

 

Mark



#19 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 16 April 2021 - 02:08 PM

Where did you find the QE figure of 68%?

 

Mark

From that same APT page https://www.astropho...phy.app/EOS.php. And I've also ran into that value (with the reference) reading forums on dpreview or PTP. As I recall, originally it was even higher.



#20 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 16 April 2021 - 02:35 PM

From that same APT page https://www.astropho...phy.app/EOS.php. And I've also ran into that value (with the reference) reading forums on dpreview or PTP. As I recall, originally it was even higher.

68% is very high and it's not clear where the APT page obtains the data.  The only original source I know of is PTP:

https://www.photonst...Canon EOS RP_14

 

It gives a figure of 56% which is still a pretty good figure.

 

Mark



#21 2ghouls

2ghouls

    Apollo

  • *****
  • Posts: 1,227
  • Joined: 19 Sep 2016

Posted 16 April 2021 - 03:45 PM

- The Ra seems to be much better than the 6D and much, much better than the Rp at lower ISO's

- The sweet spot for all of these cameras are the mid ISO ranges for DSO astrophotography (roughly 800 - 4000)

- The Ra and 6D, really appear to be better cameras for low light conditions with respect to noise.

- So signal overwhelms noise, do these read noise readings really impact image quality in day-time photography when the lower ISO would be used?

Yes, to the first three, but in practice don't obsess over read noise, it's one factor among many. Any of these three cameras can produce excellent results under a dark sky at ISO1600 with nice optics.

No to question four. Read noise will have very little impact on the image quality of a bright, sun-lit scene. 

 

 

And now a real learning question.  I have a ZWO 1600MM Pro.  I typically use it at 100% gain (default), which is something I'm starting to research just today when I accidentally found this thread.  I'm assuming 100% gain is more like high ISO (or maybe I have that backwards)?  And is it also true that there is a sweet spot on that camera, and should I back off that some to find a similar sweet spot?  So if I have a bright star in the shot, likely want to back off the gain, and shoot longer for more details?

Where are you seeing (software?) gain on the 1600MM Pro expressed as a percentage? If it were expressed as a percentage with gain 300 being 100%, I would not recommend using that gain except for perhaps narrowband. In the ASCOM driver gain is selectable at any value from 0-300 I believe. Gain is analogous to ISO, and the concerns are the same - picking a gain/ISO that balances low read noise with sufficient dynamic range. For the ASI1600MM, you could just use Gain 139 (unity gain) for everything. You could think of this as equivalent to ISO1600 on a Canon camera (this isn't at all exact, just to give you a rough idea). Personally, I use gain 75 for broadband filters (RGB) and gain 200 for narrowband (SHO). With my skies and these gains at f/5.4 I typically do 1 minute subs with gain 75 in broadband and 5 minute subs at gain 200 in narrowband.


  • kreegan99 likes this

#22 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,768
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 16 April 2021 - 04:40 PM

That's the low-light score not the low ISO score!  It's measuring the performance in low light (where the RP is good) and not the performance of the low ISO (where the RP is not good).

 

Mark

So what does “low-light ISO” mean exactly?

 

The other thing is, if I’m at ISO 100, I’m taking a daytime picture, and the sensor will be overwhelmed with light, so if it’s a bit noisy, how will I see that.  I bought mine based on Ken Rockwelll’s glowing reviews.  It’s his most used camera.  The guy knows a lot about cameras, but most of his pictures are daylight shots.  Sometimes he’ll do nightscapes, or long exposures of Yosemite.  Climbers on “half-dome”or something like that


Edited by SandyHouTex, 16 April 2021 - 04:45 PM.


#23 sharkmelley

sharkmelley

    Aurora

  • *****
  • Posts: 4,630
  • Joined: 19 Feb 2013
  • Loc: UK

Posted 16 April 2021 - 05:03 PM

So what does “low-light ISO” mean exactly?

It's defined here:

https://www.dxomark....col-and-scores/



#24 astroserge

astroserge

    Lift Off

  • *****
  • topic starter
  • Posts: 15
  • Joined: 21 Jan 2021
  • Loc: SoCal

Posted 16 April 2021 - 08:11 PM

Read noise will have very little impact on the image quality of a bright, sun-lit scene. 

 

 

I can't agree on that. The first RP reviewers (on dpreviews and elsewhere) noted significant color noise in daytime pictures at ISO 100 especially visible in "flat" areas (sky, walls etc). Though I still bought the camera. :-) 



#25 kreegan99

kreegan99

    Explorer 1

  • -----
  • Posts: 83
  • Joined: 20 May 2009
  • Loc: Endwell, NY

Posted 17 April 2021 - 07:18 AM

Yes, to the first three, but in practice don't obsess over read noise, it's one factor among many. Any of these three cameras can produce excellent results under a dark sky at ISO1600 with nice optics.

No to question four. Read noise will have very little impact on the image quality of a bright, sun-lit scene. 

 

 

Where are you seeing (software?) gain on the 1600MM Pro expressed as a percentage? If it were expressed as a percentage with gain 300 being 100%, I would not recommend using that gain except for perhaps narrowband. In the ASCOM driver gain is selectable at any value from 0-300 I believe. Gain is analogous to ISO, and the concerns are the same - picking a gain/ISO that balances low read noise with sufficient dynamic range. For the ASI1600MM, you could just use Gain 139 (unity gain) for everything. You could think of this as equivalent to ISO1600 on a Canon camera (this isn't at all exact, just to give you a rough idea). Personally, I use gain 75 for broadband filters (RGB) and gain 200 for narrowband (SHO). With my skies and these gains at f/5.4 I typically do 1 minute subs with gain 75 in broadband and 5 minute subs at gain 200 in narrowband.

Thank you very much - very helpful.

 

And thanks for identifying an assumption I had.  The ZWO X2 driver (using TheSkyX on Raspberry Pi - no ASCOM).  I assumed 100% - it may not be.  ZWO shows Gain in terms of DB on their charts for read noise.  So very likely that is the measure they are showing in their X2 driver.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: Astrophotography, DSLR, Equipment



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics