Nikon Z6/Z7 as a UV/IR Conversion: the Bad News

[DonPilou]

 

The current failure by all three of Life Pixel, Kolari Vision and MaxMax: They are all three of them offering full spectrum conversions of an unsuitable camera, the Z6 (or Z7). Apparently none of the "big 3" converters ever test a particular full spectrum camera model under all three of UV, IR and Vis. Usually it is only IR that gets tested (it seems to me).

 

BETA TESTING BY THE CONSUMER IS A VERY BAD PRACTICE FOR ANY COMPANY TO FOLLOW. This really peeves me off. WHY is it always me or Birna who gets stuck "discovering" this stuff? The D700 IR-LED shutter monitor fog. The Coastal 60/4.0 hotspot. Etc. And now the incredible worsening of the Z6 PDAF banding problem under UV filters.

 

.

 

I totally agree with you Andrea, in such a niche technique, professional companies developing dedicated gears have to test their solutions first, to only provide totally functional cameras. I was also fed up when I first received my full spectrum Canon 5D IV from Kolari, realizing there was an internal IR led leak that made it unusable for my projects. Fortunately they worked on a specific solution, and I still had my FS Canon 6D to work, but I lost 3 months.

[dabateman]

Dabateman: However, I do wonder if this is a Z6 problem or a Lifepixel problem.

Cadmium: I have been wondering if this might be some kind of LifePixel issue.

 

In Post #1 I explained what causes the Z6 banding.* To review, that would be the Z6/Z7 software algorithm which attempts to correct for PDAF striping and produces in its place PDAF banding. In an unmodified Z6/Z7, the banding is not always obvious at higher ISOs. In a converted Z6/Z7 under a UV-pass filter, it seems that the inherent banding is exacerbated to the point of unusability no matter what ISO is used.

 

I don't understand how you can wonder whether conversion, per se, as performed by Life Pixel, Kolari Vision or MaxMax can cause a banding problem because conversion involves nothing more than removal of the internal UV/IR filter, removal of shaker or shaker glass (if any), addition of a clear glass filter and an adjustment to the sensor plane as needed for DSLR infinity focus. Conversion does not involve changes to boards or to software or firmware. So conversion cannot cause PDAF banding.

 

*to be clear - I provided the explanation of banding provided by Nikon, Jim Kasson, and other camera gear gurus. There are some references in Post #1.

.

 

My guess was that Life Pixel used glass that is not good for UV. To be nice it could be an optimized AR coated glass that absorbs most of the 360nm and lower UV wavelengths, but reduces hot spots in IR. Not too far fetched as most users will be doing IR and not UV photography. You probably know all the UV photographers in the world, and may not be a large list.

 

My assumption is based on the dull UV colors Birna has shown for the Z6 compared to her other converted cameras. Possibly indicating a lot less of the less than 360nm is getting through and more of the 370 to 380nm range is getting through. Comparing to Kolari converted camera would help, but I now realize, unless they are side by side there would be no way of knowing based on white balance settings that may use in MD versus Birna on the other side of the pond.

 

Also with my recent successful Kolari conversion and great repair job performed I may be on a mental Kolari fanboy kick. They paid for shipping both ways and fixed all the problems with my recent converted camera, being the back scroll dial not responding. So I should hold off all judgement. Life pixel never responded back to my initial questions last year or my call. Whereas Dan from MaxMax (LDP) has been extremely helpful.

 

However, based on post #61 above, with horrible IR banding coming out of the Z6. I would say its save to say that this camera should not be converted.

 

On the happy side Dpreview published the high res shots using the Panasonic S1R. That camera has crazy high detail compared to 100 Mpixel medium format options. Just the file size looks huge at over 300 MB. I hope Kolari does a tear down and let us know if it suffers from IR shutter monitor problem as the other recent Panasonic cameras do.

[Andrea B.]

My guess was that Life Pixel used glass that is not good for UV.

 

Really? They've been in business for 15 years, made literally thousands of conversions,

and suddenly they are using bad glass?

:lol:

No guessing needed here. I've explained twice what causes this particular mess.

But once more with gusto: (1) Nikon's PDAF-banding Z6, and

(2) lack of testing under a UV-pass filter by Life Pixel for the full spectrum Z6 conversion.

 

 

---

 

I now have two of Birna's NEFs to send to Life Pixel. One is low-ISO and the other high-ISO. I'll post them here in a moment after I work them up to see what's going on. Birna shoots better test photos than I do, but I make better conversions. So between the two of us, we ought to be able to provide Life Pixel with a reasonable demo of this wretched banding.

[Andrea B.]

Gear [Nikon Z6 full spectrum conversion + UV-Nikkor 105/4.5 + BaaderU UV-pass filter + Sunlight]

 

As Shot [f/8 for 1.6" @ ISO-100]

This is a perfect test shot! It contains UV-white, UV-black and a known UV floral signature. It is well exposed, pushed to the right by about half a stop.

(Off Topic: Nikons have lots of headroom, so we can push to the right to get better, less noisy shadows and yet not risk color problems from pulling back the blowout regions.)

 

 

Conversion from NEF to TIF in Photo Ninja 1.3.6c

The file converted beautifully with no problems in demosaicing or white balancing. I kept it very simple. There are no shadow lifts, no sharpening, no noise tweaks, no color adjustments aside from WB. The JPG was pulled from the PN TIF in Photo Mechanic. Here are the settings.

• Demosaic NEF file: Conservative with Balance G0/G1
  
• Color correction: The dropper was dragged over the black patch on the CC card because the PTFE area was blown. In spite of this very slightly less desireable method of white balancing, the white balance is spot on.
  
  • Light source = no profile
    
  • Mode = manual
    
  • Temperature = 2000
    
  • Tint = -95
    
  • Color recovery strength = 75
    

  [*]Exposure and Detail:

  • Exposure = Absolute
    
  • Highlights = -.35 (The PTFE was blown out by about 1/2 stop.)
    
  • Detail = 15 (Between 10-15 is normal for UV work.)
    
  • NO other slider tweaks in this tool.
    

  [*]Color Enhancement: Plain/65 (My usual choice for UV.)

  [*]No noise reduction or sharpening was applied.

The false colours are perfect here: yellow, blue, white, black, grey. I am hoping the JPG crunch for posting here has not shifted them too much.

 

 

Raw Digger Raw Composite

Nothing surprising about the raw color appearance. The red channel is brightest, followed by blue, then green. This gives a typical color cast somewhere between magenta and purple.

 

 

Raw Digger Histogram

Now I'm a bit surprised! Don't you think that the histogram looks a bit "comby"?? I ignored the red channel blowout spike and adjusted the y-axis to fit the non-blowout peak to emphasize the combing.

 

 

Raw Digger RGB Render with no auto exposure push.

Looks good. RD doesn't push the raw color sat in case you were wondering why it's a bit under.

 

 

Screen shots from Photo Ninja at 100% to show the banding

I made these 3 screen shots from an even more conservatively converted version of the file. Instead of using the Absolute setting, I used the Unadjusted setting in Photo Ninja. I wanted to prove the point about the banding by using a conversion which didn't reset the white point and thus induce a pseudo-shadow lift.

 

Everything was looking so good until we got to this !!! Let's put it this way: you could not successfully print this file. Nor could you successfully sell it for printing or display in a magazine or a museum. It is useless.

Dang it. This is so sad. It would be just the COOLEST camera ever if not for this miserable banding mess.

 

Close your bookmarks, fully expand your browser, click the photo and then click again if you see a + icon.

 

 

EDIT: I am replacing the 100% zooms with better versions.

Now done. 23 Apr 2019 7:09 PM EST (US)

[Andrea B.]

This is the ISO-1600 test shot.

To review: Nikon Z6 full spectrum conversion + UV-Nikkor 105/4.5 + BaaderU UV-pass filter + Sunlight.

Exp: f/8 for 1/8" @ ISO-1600

Conversion in Photo Ninja using Absolute setting as described above.

 

Comment: The noise is so very fine grained that it hardly seems worth the effort to denoise?

I do note that what noise there is serves to slightly mask some of the banding in some areas,

but not in all. Indeed, most of the banding is still quite obvious.

 

 

 

Here are some 100% crops from this ISO-1600 version.

[Andrea B.]

Alrighty then...that is all I can do for now. I have jury duty for the remainder of the week, so I'll be back next week to send this stuff to Life Pixel. I've already emailed them about the Z6 problem, but also want to supply them with some hi-ISO and lo-ISO photos.

[Cadmium]

I suppose you could have the camera converted back to visual. I don't suppose any conversion place does a reverse conversion for a reduced price if the expected results are yet unknown for that camera model.

That would be a nice thing.

[nfoto]

What annoys me the most is that we are de facto Alpha testers. Not beta.

 

If the conversion firms really had tested their modified cameras at all, they would have seen what the poor customer easily observes: a camera that is not suitable for modification.

[Alaun]

Yes, but we should also keep in mind, it is a difference, whether we ask for a conversion or ask for converted camera. :)

[Cadmium]

It is not like they just convert your Z6/Z7, they also sell new Z6 and Z7 cameras converted off the shelf. Not a good idea if there is an inherent problem with the model like this.

 

A brand new Z6 converted

https://www.lifepixe...mera-conversion

 

A brand new Z7 converted

https://www.lifepixe...mera-conversion

[Cadmium]

Just curious, not that it should really matter, but did you, Andrea and Birna, send in a camera for conversion or buy one of their new ones for conversion, maybe you already said somewhere, but I didn't find it.

[nfoto]

I purchased a brand new camera including its conversion directly from LifePixel. That is the major cause of my annoyance with this company.

[Cadmium]

Birna, Ask for a full refund, including return and taxes. It is up to them if they want to convert such a camera, but no one should have to pay anything for something that is not useful.

They must be selling these, since they have several packaged options listed on their page. I can only imagine there are many people who are experiencing the same dissatisfaction.

It is also sad that this might be the direction of some future Nikon (and other) cameras.

It is good you tested this and defined this issue so others will know better.

[nfoto]

Some sources eg. https://www.dpreview...s/post/61856758 mention banding is less obvious or even gone when using 12-bits RAW files. I have tested this and in the case of my bad Z6, the issue is there no matter what bit depth is used.

[Cadmium]

Birna, Any update or resolution on this? Are you going to keep it or send it back?

[nfoto]

We are awaiting the response from LifePixel.

[Andrea B.]

I sent Life Pixel the sample photos. Reply from Daniel Malkin was that we are seeing the "pattern of the actual sensor". Although I am not quite sure what that means. I have asked him how to return the camera (subject to how much it has already been used).

[Andy Perrin]

Reply from Daniel Malkin was that we are seeing the "pattern of the actual sensor". Although I am not quite sure what that means.

It sounds like he's saying that he thinks it's the read lines of the sensor, as opposed to being the PDAF banding issue.

 

See the response here:

https://photo.stacke...s-banding-noise

[Andrea B.]

It really doesn't look like read-out banding to me. But I am just guessing. I take the word of Kasson, et. al., who do so much analysis of read-outs, noise, banding and so forth.

[Andy Perrin]

Have you tried asking Kasson et al. what they think of this particular noise? They might be able to tell you how to distinguish it from the usual sensor banding noise.

[Andrea B.]

Given that the converted Z6 banding matches the banding which they are showing and describing to be a result of the fix for PDAF striping, I haven't yet ask them anything about its exacerbation under a UV-pass filter. But I might write to Jim next week about this.**

 

Even if we know what causes the banding, that does not fix the problem with using this converted Z6 as a reflected UV photography tool. It would be immensely cool if Nikon itself attempted to fix this banding even though it occurs only occasionally in an unconverted Z.

 

 

[**I am yet-once-again-busy this weekend with "social duties", so I can't pursue the banding details just now. My SigOth runs a chamber music society which is presenting a world premiere of a new work this Sunday so there is much to prepare for. And there *will* be Cake! If all the alternate events in my life continue at such a fast pace as has happened recently, then I'm going to have to get someone to manage UVP, at least temporarily. It has been frantic around here.]

[Cadmium]

You might want to ask the other Dan at MaxMax what he thinks of the lines.

You can even call him voice, not that you can't call Dan at LifePixel voice, I have before, but Dan at MaxMax will defiantly tell you what he thinks, and if he thinks it is normal for that camera, and if there is any way to fix that.

You could call Kolari also. I think those three places are the big three in USA right now, maybe.

Next time I get any conversion, I am calling whoever first, and getting a guarantee that they will refund me in full if I am not satisfied with the results, or fix the camera if it is their fault.

I can't afford to buy cameras, have then converted, and be disappointed with the results. This all makes me hesitant about getting a new conversion.

I think LifePixel should give you a full refund, not based on the time you used it. After all, you and Birna are the ones who discovered this problem, and your time, and your disappointment. You didn't get anything out of this camera.

 

I feel a bit embarrassed about saying this, but I have a D7200 which I had converted by LifePixel years back now.

Full spectrum and visual filter work good, but put any UV or dualband or IR longpass on it and the shots are all fuzzy and blurry, looks terrible.

Why? I have no clue. Called Dan, he said send it in with a lens...

Well I never did, I put it in the cupboard, and it just sits there. I simply put it off, didn't have time, excuses...

I simply don't have time for that sort of rigamarole.

I bought that D7200 new from B&H, it probably has 30 or so actuation. Maybe more, but not many, I only got it to convert it.

I know that is a completely different camera, situation, problem, and I know I should have resolved that back then,

but this is partly why I am hesitant.

 

Another question I have about LifePixel, and all other camera conversion places when they sell 'new cameras' that are converted.

What are these 'new cameras', where do they get them? Are they gray market? Are they refurbished? Are we really getting the same new camera we would if we bought one new?

Is there a way to check serial numbers of cameras to find out if they have a history of refurbishment, or if they are really new?

Just wondering is all, just hesitant about buying a new camera that is converted rather than buying a new camera from somewhere else and sending it in.

If anyone has any ideas about that...?

[Andrea B.]

What are these 'new cameras', where do they get them? Are they gray market? Are they refurbished? Are we really getting the same new camera we would if we bought one new?

 

Yes, they are new US guaranteed cameras. Kolari, MaxMax and LifePixel are all legit. They are not going to sell you a grey market, used camera. All the right papers arrive in the box with the camera. And you can always check the actuation count. There will be a few because they adjust the focal plane and test that. And they usually add a white balance adjustment which would account for a few more.

 

Usually a conversion shop simply orders the new camera from B&H (or whomever) just like you would. I've bought new camera conversions from all three of MaxMax, LifePixel and Kolari. Some of them might now have a reseller's license for some camera brand, but I doubt it because that's a lot of work to accumulate licenses for so many different brands.

 

Given that any warranty is null & void once a conversion has been made, then some of your fears are unwarranted (so to speak). La! Sorry couldn't resist the wordplay.

 

 

---

 

You do sometimes need to return a DSLR conversion for a minor adjustment of the focus plane. There is a standardized focal length against which the adjustment is made in IR. So you can a hit lens + filter combo for UV which does not quite work well with the initial adjustment. If manual focus does not compensate for such a minor mal-adjustment, then you need to send in the lens with your camera for a more precise adjustment. It is not necessarily the fault of the conversion shop because they cannot guess what you are going to be using.

 

 

.

[nfoto]

The problem "child" has been put aside for a while as I enjoyed the Scottish airs for some time. I did update my stock Z6 with the new v.2 Firmware whilst abroad and found several more or less obvious improvements of this upgrade. I immediately implemented the v2 to my bad Lifepixel Z6 and there seems to be some, although not earth shaking, improvement in the banding issue. Photo Ninja still struggles whilst RawTherapee by virtue of its clever PDAF filter option in fact more or less eliminates the striping. However, RawTherapee really goes haywire if one tries to use any 'click-white' operation on the UV captures by the Z6, thus one has to coax an useful w/b by alternative, devious schemes inside the software. I'm torn between a rock and a hard place here :(

 

The superior ISO performance of the Z6 easily allows casual, hand-held shooting even with the 105/4.5 UV-Nikkor. Now, that opens new vistas on its own.

 

(UV-Nikkor, Baader U, f/5.6, 1/100 sec, 6400 ISO)

 

(Fisheye-Nikkor 8-15mm f/3.5-4.5, Wratten 87C gel in rear filter slot, f/16, 1/20 sec, ISO 250)

 

There is a weakly defined hot spot with this lens. Usually it is possible to "eliminate" this issue by judicious placement of the camera. I do wish the issue hadn't manifested itself in the first place, though.

[nfoto]

Another IR capture with the 8-15 Fisheye-Nikkor and the modified Z6. The hot spot tendency is pretty obvious here.

 

 

In a medium-low contrast IR scene like this, any banding is well hidden and of no immediate consequence.