18 replies to this topic

[DanielJStein]

Welp, guess I will be the guinea pig for this one. I just got my Nikon Z6III today and my inattention is to mod it and use it solely for astrophotography. I am mainly a landscape/Milky Way widefield kind of guy, and have used my astro modded Z6 for almost 6 years now. I figured it was time for an upgrade. I have used the Z8 and the star view mode is selling feature in it of itself, so when the Z6III listed it as a feature, I bit the bullet.

 

I am wondering if any of the brain surgeons here can point me in the right direction to sending them certain files for analyzation. Curious on things like dynamic range, ISO invariance, dual gain, and spatial filtering. Fingers crossed!

[sharkmelley]

Dynamic range, ISO invariance and dual gain are all determined by the read noise at each ISO.

 

Useful test frames are as follows:

• For each ISO, a pair of bias frames and a pair of flats is required in order to calculate the read noise at that ISO.
• Analysis of spatial filtering requires a single ISO 1600 long exposure dark at room temperature - anything 30sec or more will be adequate.
• To analyse hardcoded corrections that potentially give rise to coloured concentric rings/waves then take a series of ISO 100 flats according to my test protocol:  https://www.markshel...ml#TestProtocol

I'm happy to analyse the files if you upload them to file-sharing site.

 

Mark

[DanielJStein]

Dynamic range, ISO invariance and dual gain are all determined by the read noise at each ISO.

 

Useful test frames are as follows:

• For each ISO, a pair of bias frames and a pair of flats is required in order to calculate the read noise at that ISO.
• Analysis of spatial filtering requires a single ISO 1600 long exposure dark at room temperature - anything 30sec or more will be adequate.
• To analyse hardcoded corrections that potentially give rise to coloured concentric rings/waves then take a series of ISO 100 flats according to my test protocol:  https://www.markshel...ml#TestProtocol

I'm happy to analyse the files if you upload them to file-sharing site.

 

Mark

 

Thanks Mark, I will get you these files today via Dropbox. Just a few questions before I start:

 

For the read noise test, does it matter which lens and aperture I use to shoot the flats for this test? I will also be using just a flats panel to complete, hope that is okay as well. I will be exposing to center the histogram, let me know if that will work. Lastly, do you need each 1/3 stop of ISO or are full stops ok? 

 

Thanks for the help.

[sharkmelley]

It's best to shoot with a lens and aperture with the least vignetting and switch off all lens corrections.  Flat panel is OK as long as it doesn't produce horizontal banding from the refresh frequency.  The full stop ISOs are fine unless you are also desperate to know what happens at intermediate ISOs.

[nhmorgan79]

Interested to see how this performs. I used a modded Z7 and Z6ii for a long time. I suspect, however, that the new sensor is going to come with a little bit of a DR and noise penalty from the previous sensor.

[DanielJStein]

It's best to shoot with a lens and aperture with the least vignetting and switch off all lens corrections.  Flat panel is OK as long as it doesn't produce horizontal banding from the refresh frequency.  The full stop ISOs are fine unless you are also desperate to know what happens at intermediate ISOs.

Thanks Mark, I'm on it. Today is a little cloudy actually so I may be able to do white t-shirt method, eliminated flicker from the mix here.

[DanielJStein]

It's best to shoot with a lens and aperture with the least vignetting and switch off all lens corrections.  Flat panel is OK as long as it doesn't produce horizontal banding from the refresh frequency.  The full stop ISOs are fine unless you are also desperate to know what happens at intermediate ISOs.

Files are attached here. Please let me know if I did this properly. I ended having to use the flats panel after all.  

[sharkmelley]

Files are attached here. Please let me know if I did this properly. I ended having to use the flats panel after all.  

They look fine.  I'll work through them and report back.

[sharkmelley]

First off, here are the read noise results:

 

 

I used the green channel for these calculations because this is not affected by the digital scaling applied to the red and blue channels.  The read noise figures are much higher than I expected - approx 50% higher than the Z6 and Z6II, with a corresponding loss in dynamic range.  High conversion gain kicks in at ISO 800, just like the Z6 and Z6II.

 

Here are Bill Claff's Z6 and Z6II read noise charts for comparison:

https://www.photonst...,Nikon Z 6II_14

 

Perhaps the higher read noise is the downside of a partially stacked sensor?

 

 

[sharkmelley]

Here's the plot to discover whether or not spatial filtering is taking place:

 

 

Noise reduction is switched off but clearly there is spatial filtering because the entire left hand vertical arm does not exist.  This missing arm represents pixels that are much brighter than their 24 neighbours.  It's difficult to tell much else without a much longer exposure.

 

However, look at the raw data filtering section of my Z6 review here:

• https://www.markshel...nz6_review.html

Although there was strong evidence of spatial filtering, there was no observable damage to small tightly focused stars and also calibration with dark frames still worked well.  So this indicates that spatial filtering is not necessarily destructive.  The same is likely to be true of the Z6III.

[sharkmelley]

Finally, the results of the tests for a hardcoded correction with an attached Nikkor Z 50mm f1.8 S  lens:

 

 

 

Lens corrections were switched off but just like previous Nikon mirrorless cameras some kind of correction (probably for colour shading) is taking place.

The top row shows the (bias-subtracted) red channel divided by the (bias-subtracted) green channel.

The bottom row shows the (bias-subtracted) blue channel divided by the (bias-subtracted) green channel.

Note the progression of the rings from left to right as the exposure level drops by one stop each time.

 

For landscape/Milky Way widefield astrophotography this correction should not prove to be a problem but coloured concentric rings could become visible when imaging deep-sky objects where many exposures are calibrated, stacked, light pollution subtracted and then stretched.  In that case, the best workaround is to use a high ISO and expose so that the peak of the back-of-camera histogram is near the centre.

[DanielJStein]

Wow Mark, thanks for the analysis! Unfortunately your results are about what I expected due to the use of the partially stacked sensor.

I was hopeful that the high conversion gain would be a little better than the predecessor z6, especially given the higher native ISO range. Glad to hear spatial filtering shouldn’t be an issue though, but also sad to hear we will still need to work around concentric rings.

Hopefully I can get some real world results next week during new moon. I mainly shoot Milky Way/widefield anyway. I like having the star view mode, but almost wondering if I should
Keep my z6 HA mod around for a little longer.

[photobookie]

How do these results compare to a Nikon Z8?

[sharkmelley]

How do these results compare to a Nikon Z8?

Z8 read noise can be found at PhotonsToPhotos:

• https://www.photonst...14,Nikon Z 8_14

 

Diagnosis of Z8 concentric rings caused by the hardcoded correction can be found here:

• https://www.cloudyni...ngs/?p=13420852

 

As for Z8 spatial filtering, no-one has sent me a long exposure dark for analysis.

[DanielJStein]

Z8 read noise can be found at PhotonsToPhotos:

• https://www.photonst...14,Nikon Z 8_14

 

Diagnosis of Z8 concentric rings caused by the hardcoded correction can be found here:

• https://www.cloudyni...ngs/?p=13420852

 

As for Z8 spatial filtering, no-one has sent me a long exposure dark for analysis.

I might have a Z8 dark. I do not own one but I have had several loaners.

 

Update: I do not sadly...

Edited by DanielJStein, 01 July 2024 - 12:39 PM.

[sharkmelley]

Bill Claff has just posted his Nikon Z6 III results:

• https://www.photonst...Nikon Z 6III_14

 

The results are similar (but not identical) to the ones I calculated earlier and it confirms that the Nikon Z6III has higher read noise than the Z6 and Z6II.

[DanielJStein]

Bill Claff has just posted his Nikon Z6 III results:

• https://www.photonst...Nikon Z 6III_14

 

The results are similar (but not identical) to the ones I calculated earlier and it confirms that the Nikon Z6III has higher read noise than the Z6 and Z6II.

I just saw that the other day as well. I think I'll stick to my modded Z6I for astro and use the Z6III for daytime photography such as wildlife and landscapes. The AF is really good.

[DanielJStein]

Welp, did my first light of astro with the Z6III. I do think my legacy Z6 I might have a slight edge as previously discussed. Either way, I am still happy with how this shot came out. Hiked up a mountain in my favorite place (Adirondacks) with 2 other bodies, 4 lenses, and my star tracker. This is just one of many from that night. Z6III w/ Z-70-200 2.8 S @135mm. 190 30” images @ f/4 manually stitched in PS.

 

I do enjoy composing and focusing with the new body over the old however. The star view mode works great, and even the AF will focus on a brighter star.

Attached Thumbnails

• 

Edited by DanielJStein, 14 July 2024 - 01:59 PM.

[Alen K]

Welp, did my first light of astro with the Z6III. I do think my legacy Z6 I might have a slight edge as previously discussed. Either way, I am still happy with how this shot came out. Hiked up a mountain in my favorite place (Adirondacks) with 2 other bodies, 4 lenses, and my star tracker. This is just one of many from that night. Z6III w/ Z-70-200 2.8 S @135mm. 190 30” images @ f/4 manually stitched in PS.

I’d say your tracker is malfunctioning. .   

 

https://www.youtube....h?v=g-4-gLlF0uw