Jump to content

SkySurveyBanner.jpg.21855908fce40597655603b6c9af720d.jpg

Thermal noise evaluation of 3 Nikon bodies

chrisvdberge

By chrisvdberge
in Discussions - Cameras

Recommended Posts

chrisvdberge

chrisvdberge

Posted
Posted

I'm doing a comparison of the thermal noise levels with the Nikon D7000, D5100 and the D600. 

I expected to get similar results for the D7000 and the D5100 since the chip is exactly the same. But I get a surprising result:

post-42588-0-82644100-1425130334.png

I took 5min exposures at ISO800, same location and room temp for all 3 if them, same settings (all noise reduction OFF) and I did the run of 30x5min for the Nikon D7000 twice to confirm the values. (2nd time they were even lower). 

RAW files are loaded in PixInsight to read the mean values (all signal is noise in a dark frame)

Am I doing something wrong or missing something, or is the D7000 somehow an excellent performer in terms of thermal noise. And if so, how/why?

Any insights?

Link to comment
Share on other sites
  • Replies 26
  • Created
  • Last Reply
StuartJPP

StuartJPP

Posted
Posted

I am not too familiar with Nikon cameras but don't you have to load hacked software to get true RAW files?

I was under the impression that Nikon adjusted the black point, especially on older cameras....

Though you could ignore me as I said I am not too sure with Nikon.

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

That's an interesting point. I'm not 100% sure, but I think all camera's (might) do some change to the signal already in the analog phase of the signal chain (before the ADC's). This could have been changed indeed, but then I am surprised about the difference between the D600 and the D7000 since the D600 is a newer model. 

Furthermore, the 'rate of change' over the # of exposures is also quite different:

post-42588-0-72918100-1425133430.png
Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

Yes, they fiddle the blackpoint and remove single pixels with high values, the latter being very damaging to faibt stars that occupy single pixels.

/p

I guess you are referring to the 'star eating algorithm', which is something that is gone for over 5 years already. My Nikon D70s has it, but no problems at all with my D7000. 

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

I just realized that the D600 has more pixels, so probably this is explaining the difference between the D7000 and the D600? (assuming that more pixels means more thermal noise?)

although there are fewer pixels occupying the same area because the D600 is full frame.....

(is there no way to edit a post? ;))

Link to comment
Share on other sites
Mr Spock

Mr Spock

Posted
Posted
I guess you are referring to the 'star eating algorithm', which is something that is gone for over 5 years already.

Indeed. It is that kind of misinformation which could put Nikon owners off trying AP ;)

You can edit posts once you have reached 250 posts :)

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

I have a D7100, how did you carry out the test ... Shall I do it on mine ?

That would be great! 

I closed of the viewfinder with some tape and put the body on it's back. All noise reduction settings OFF (long exposure, high ISO), preview OFF

ISO 800 and shooting RAW, 30x 300sec. with 3sec interval. Room temperature (around 20degrees celcius)

I measured the mean value of the frames in PixInsight with the Statistics procedure. (per channel.. added them up for the graph above. )

If you don't have PixInsight maybe you could share the files?

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

Indeed. It is that kind of misinformation which could put Nikon owners off trying AP ;)

You can edit posts once you have reached 250 posts :)

Indeed.... while Nikon is outperforming Canon in terms of sensor performance for years now ... ;) 

Need to have some patience then for hitting 250 posts ;)  (and think before I hit 'Post' ;) )

Link to comment
Share on other sites
knobby

knobby

Posted
Posted

Ok, I'll do this again tomorrow ( With fully charged batteries!)

here are the first 9 to give you an idea ...

dont forget the D7100 is 24 meg so the files are large

https://drive.google.com/folderview?id=0B2QNGXiAfO_RTkRBNDF4MHlfOUU&usp=sharing

uploading now.

I might try this test again with the camera in crop mode, could be interesting as the sensor seems to glow at the edges

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

Nice, thx for participating :) 

Running discussion on Facebook as well and got the insight that apparently PixInsight is doing something on the RAWs, so the measurement is not as correct as I would want to. 

Also there is still HPS (HotPixelSurpressing) and lossy2 compression active in the Nikon Firmware, which might differ per model. So this could also explain the differences. 

I'll try to patch my D5100 with the adjusted firmware and see what effect it has on this measurement :) 

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

That greatly depends on the software you use to view the raw and to convert it. Some scaling will be done on opening and sometimes stretching as well. Definitely for converting (like yours... definitely stretched ;)) That makes it difficult to get 'true readings' or comparisons ;)

Link to comment
Share on other sites
perfrej

perfrej

Posted
Posted

I guess you are referring to the 'star eating algorithm', which is something that is gone for over 5 years already. My Nikon D70s has it, but no problems at all with my D7000. 

Hmm... I was under the impression that some bad form of it still existed. I'll do some digging. I have a D3s, by the way. Maybe I should use it for astro as one of my CCDs is flaky...

/per

Link to comment
Share on other sites
chrisvdberge

chrisvdberge

Posted
  • Author
Posted

Hmm... I was under the impression that some bad form of it still existed. I'll do some digging. I have a D3s, by the way. Maybe I should use it for astro as one of my CCDs is flaky...

/per

The HPS (hotpixel surpression) is indeed still active, but not so much affecting your lights by eating stars, but more affecting the bias and darkframes so they are less effective. 

Will try to patch my D5100 with other firmware today to see the difference

Link to comment
Share on other sites
sharkmelley

sharkmelley

Posted
Posted

Hmm... I was under the impression that some bad form of it still existed. I'll do some digging.

This gives the lowdown on the algorithm used in the D5000 and (I believe) subsequent models:

http://www.dpreview.com/forums/post/37071846

Accurately measuring the dark current is fraught with problems because of the Nikon black clipping issue.  However, with quite a bit of maths I ended up with a result for the D7000 which I'm reasonably confident with.  More info in my Cloudy Nights thread on the subject:

http://www.cloudynights.com/topic/408187-nikon-d7000-read-noise-gain-and-thermal-noise/

Mark

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.