Jump to content

Stargazers Lounge Uses Cookies

Like most websites, SGL uses cookies in order to deliver a secure, personalised service, to provide social media functions and to analyse our traffic. Continued use of SGL indicates your acceptance of our cookie policy.

sgl_imaging_challenge_banner_beauty_night_skies.thumb.jpg.2711ade15e31d01524e7dc52d15c4217.jpg

MikeODay

Nikon D7500 -v- D5300 - comparison of read noise and patterns in bias frames

Recommended Posts

I have completed a comparison of the level and pattern of read noise in bias frames between the Nikon D7500 and the D5300, please see the blog page ...

 

 

Edited by MikeODay

Share this post


Link to post
Share on other sites

Does that noise control mean that when taking lights, particularly in bulb mode, that the fainter stars are removed?

Share this post


Link to post
Share on other sites
51 minutes ago, happy-kat said:

Does that noise control mean that when taking lights, particularly in bulb mode, that the fainter stars are removed?

I don’t believe so.  In fact I have no reason to believe that noise reduction processing is being carried out in the camera.  

Like the D5300, the camera added bias is retained in the NEF files ( in this case it is a 400 ADU pedestal -v- 600 for the D5300 ) and the noise is either side of this value so I think there is no clipping of the lowest signal levels.

Share this post


Link to post
Share on other sites

I have now added a set of results comparing the noise in dark frames taken with the D7500 and D5300 with exposures ranging from 1 sec to 240 seconds...

 

Share this post


Link to post
Share on other sites

That's an interesting analysis.  I especially like the decrease in glow at the bottom edge of the sensor in the bias frames.  It indicates that Sony is starting to address this issue.

In your dark frames I don't understand why the top D7500 dark frame (which I assume to be the 1sec exposure although you say 1sec is at the bottom) appears so much brighter than the top D5300 dark frame.  Are these in-camera jpegs that you have stretched by identical amounts or are they raw files that you have treated somehow?  It's not clear exactly what we are looking at.

Also interesting is how the overall hue of the dark frames varies between purple and green as the exposure increases.  That intrigues me - I wonder what causes it?

Mark 

Edited by sharkmelley

Share this post


Link to post
Share on other sites
20 minutes ago, sharkmelley said:

That's an interesting analysis.  I especially like the decrease in glow at the bottom edge of the sensor in the bias frames.  It indicates that Sony is starting to address this issue.

In your dark frames I don't understand why the top D7500 dark frame (which I assume to be the 1sec exposure although you say 1sec is at the bottom) appears so much brighter than the top D5300 dark frame.  Are these in-camera jpegs that you have stretched by identical amounts or are they raw files that you have treated somehow?  It's not clear exactly what we are looking at.

Also interesting is how the overall hue of the dark frames varies between purple and green as the exposure increases.  That intrigues me - I wonder what causes it?

Mark 

Hi Mark

The order of the images is, from the top, 240, 120, 60, 30, 16, 8, 4, 2, 1 seconds.

Capture:  14 bit NEF 

Processing: 

- no calibration ( of course :) )
- debayer
- subtract the difference in the mean level in the images produced by the two cameras ( 400 for D7500 -v- 600 for D5300 ) 
- autostretch the D7500 1 second image and, using the screenTransferFunction, apply this stretch to all of the images
- reduce in size
- capture the screen and cut and paste the individual frames to produce the sets of images above

As I was typing this I realise that I did not alter the default interpolation algorithm used for downsizing so it may have not been the best choice for such an extreme reduction ( ~ 6000 down to 400 ).  This process may have contributed to the variation in hue.  I will need to look into it.  

I was also surprised by the apparent difference in the brightness of the D7500 240 sec image.  The mean levels are all the same and the std Dev of the 240s is only slightly higher than that of the 120 sec image - so why the difference?  What I did not do was re-examine the mean and std dev after downsizing - perhaps a difference was introduced in that step.  

Cheers

Mike

 

Share this post


Link to post
Share on other sites

My mistake!  Of course the longest exposure darks will be the noisiest.  It might also explain the difference in brightness.  For a long exposure dark, it is more difficult for the camera to "judge" the mean level of the accumulated dark current that needs to be subtracted.  Sometimes it is slightly off.  So the overall brightness is not important - it is something that does vary slightly from model to model.

Mark

Share this post


Link to post
Share on other sites
10 hours ago, sharkmelley said:

My mistake!  Of course the longest exposure darks will be the noisiest.  It might also explain the difference in brightness.  For a long exposure dark, it is more difficult for the camera to "judge" the mean level of the accumulated dark current that needs to be subtracted.  Sometimes it is slightly off.  So the overall brightness is not important - it is something that does vary slightly from model to model.

Mark

Ok some more info...

Both the apparent difference in hue and brightness is real and present in the full size images.

An examination of the measured noise in each channel I believe may explain the apparent shift in hue.

1 second image:

R:    1.393.  Or 101.5% cf. green
G:    1.373.  
B:     1.393.  101.5% 

240 s image:

R:  5.900.  Or 111.3% cf. green
G:  5.302
B:  5.875.  110.8%

So the standard deviation of the noise in the red and blue channels is increasing more quickly than that in the green.  The result is that the brightest pixels in the red and blue channel are brighter than those in the green channel and hence I believe the apparent ( but not the mean ) hue shifts away from green.  

Likewise I think that the the overall apparent brightness of the 240 sec image is due the greater std dev of the noise.  That is, the std dev of the noise is 50% greater than that of the 120 sec image and hence it appears brighter.

Does that make sense?

Now why does the std dev increase less for the green channel?  The std dev of the read noise is essentially the same in all channels.  So I guess the difference is due to the thermal noise and perhaps is due to the fact that there are twice as many green pixels.

Edited by MikeODay

Share this post


Link to post
Share on other sites
21 minutes ago, MikeODay said:

So I guess the difference is due to the thermal noise and perhaps is due to the fact that there are twice as many green pixels.

Ah, no...

I have just checked the per channel noise in the CFA file and the difference is there as well;  std dev of green channels about the same and both less than either of the red or green.

Does anyone know why that might be ?

Share this post


Link to post
Share on other sites
10 hours ago, MikeODay said:

I have just checked the per channel noise in the CFA file and the difference is there as well;  std dev of green channels about the same and both less than either of the red or green.

Does anyone know why that might be ?

Most Nikon cameras digitally scale their red and blue channels up or down.  You can check this by looking for regular histogram gaps or peaks.  A factor of 1.1 or 0.9 sounds about right.

Mark

Edited by sharkmelley

Share this post


Link to post
Share on other sites

Here's a typical example of the red/blue channel scaling - the histogram gaps provide the clue.  This example is from a Nikon D800.

Mark

NikonD800HistogramGaps.thumb.jpg.602ff2e4f75e6ef960b03876176349f4.jpg

 

Share this post


Link to post
Share on other sites
On 02/11/2017 at 10:22, sharkmelley said:

Here's a typical example of the red/blue channel scaling - the histogram gaps provide the clue.  This example is from a Nikon D800.

Mark

 

 

Thanks Mark.

So am I correct in thinking that this scaling is independent of white balancing?

That is, the scaling is applied to the raw channels and if one wanted to duplicate the in-camera white balance ( that is, the camera would produce in an in-camera jpeg ) one would apply the camera white balance factors reported by Pixinsight to the post-scaled raw image.  The alternative I guess is that one would have to discount the white balance factors by the channel scaling factors.

Share this post


Link to post
Share on other sites

The camera white balance factors already take into account this red/blue pre-scaling of data. 

Mark

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.

  • Similar Content

    • By R26 oldtimer
      I see that there are some great actions for PS out there but they come at a price. Have you come across any free and good actions, that you can recommend?
    • By bottletopburly
      Having upgraded from canon 1000D to Nikon D5300  how do users of the 5300 find it for Astro I will be using with APT , and how do you take flats as I can’t use Av mode as there is nothing in tools menu to choose no lens ,  do you use in manual mode  and look at histogram ? .
    • By masjstovel
      Hi,
      I bought my first telescope, SW 150pds about 6 months ago with the purpose of astroimaging "when i feel ready".  So far ive used my Nikon D810 for that, and I'm now  planning on taking the step buying my first AP camera.
      My targets would be DSO's, and not planetary. I want a mono-camera, not color. 

      I want to get away with a very good camera to a reasonable price (wouldn't we all...) and in this regard I've been drooling on the ZWO ASI 1600MM Mono for some time. The price for it is in the upper part of my budget, but I'm willing to if its worth it.  I've seen from other treads that sensor-size isn't everything, and dynamic range and gain and all is just as important, but i have trouble understanding it all 100% when it's all new to me, but in my experience i am a practical person who learns things much better and faster with the gear in my hand. So without getting to technical, and staying as objective as possible - please help me with; 

      1. Is this a good camera to go for?
      2. It's sold with options of filters 1.25", 31mm or 36mm - Why these options, and what determines what i would choose?

      3. Would you go for another camera in this price range, and why? - Or to rephrase it a bit; If you were in my shoes, which camera would you og for?

      I'd appreciate any help:)

      I might add, that i understand that with my lack of experience, buying a mono-camera with filters and all might seem premature, but for some strange reason. I enjoy these "way over my head"-projects and figuring  out things as time goes - I just need some guiding in the right direction. 
    • By jambouk
      We are running a session at my local society on transits and occultations. One station will focus on exoplanet transits, and we'd like to build a very simple model to demonstrate this. We have a star (light source) and an orbiting "planet" but I need to work out how to detect the changes in light intensity and display this on a laptop, like a classical transit photometry trace below (taken from https://heasarc.gsfc.nasa.gov/docs/tess/primary-science.html).
      Is there a way to take a feed from a DSLR through the USB output to do this, else I could get an adapter for my ZWO and put an EOS lens on the front of that. I really do want a light intensity vs time trace in real time on the laptop. This model will be run in a darkened room.
      Thanks for any comments.
      James
       

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.