Canon EOS 6D B&W Conversion at High ISO

[StephanN]

In a recent thread about possible harm to insects when flashing them (https://www.ultravio...cts-with-flash/), naturally the question of using high ISO popped up. ISO-performance has been mentioned all across the board in different topics, the merits of different companies and models discussed, etc. (and is a constant source of heated discussion all over the internet), but Canon-users on this site are few and far between. So I wanted to test my UV-workhorse Canon and see how it performs at high ISO:

 

The camera: Canon EOS 6D, introduced in 2013 (https://en.wikipedia...ki/Canon_EOS_6D), a full-frame camera with 20.2 MP, native ISO range from 100 to 25600, with the option of going to 50 at the lower end, and 51200 and 102400 at the upper end, respectively. Now, 8 years later, I would expect new models to show a significant improvement at higher ISO, as is the case with all manufacturers and lines. Its successor, the 6D II (2017) already sports a native ISO range from 100 to 40000 (and the latest mirror-lens bodies 100 to 102400, so things are developing). I bought it actually for the very reason that being the discontinued model it was cheaper to be had second-hand

 

It was converted by maxmax to black & white, i.e. getting rid of the Bayer-filter, and at the same time removing the usual blocking filter, replacing it with their proprietary X-Nite 330C-filter. This filter does have a small IR-leak of 1% at 720nm (https://maxmax.com/s...r-x-2-2mm-thick ). Converting the camera to b/w boosted its UV-response by an unknown factor (I've read claims of fivefold somewhere), and as a matter of fact, in most cases the difference of camera and camera+S8612-filter is all but non-existent.

 

So, before I did this test, I've always tried to keep ISO below 1600. If I shoot with tripod, then ISO can be as low as 100 (for perfect quality), but if I don't want to smear out movement or want to take a handheld shot, then up goes the shutter speed, and up goes ISO as well. The prospect of shooting handheld was my main motivation for getting this modification; especially when I'm on city-trips and with some friends I want to be able to take snap-shots.

 

So, without further ado, if you've braved this somewhat lengthy introduction, here are the results.

 

I've shot a stamp of mine, indoors, with camera on the tripod, distance sensor to stamp about 55cm (I think I missed perfect focus ever so slightly). The lens was the EL-Nikkor 105 mm f/5.6, set to f/8, used on a helicoid, of course. A full-spectrum Yongnuo VN560III was triggered by radio from the camera. I shot a series from 100 to 25600, trying first to adjust the level of flash so that the initial histrogram would be about the same, and for 25600 I shot some more tests where I allowed some under-exposure, up to about 3 stops. I didn't think it necessary to go lower. For 51200 and 102400 I just tried to keep the histogram around the centre. I'm not showing all of them, this would be too boring even for myself

 

This is a 1000x1000 px crop from the 5472x3648 pixel frame. In post (Lightroom) I mainly adjusted brightness, structure and clarity. Noise reduction was only performed from 25600 upwards. The stamp has a size of about 24x29 mm (on top you can see the technical paper I placed it on, the thick lines are spaced 0.5 cm apart).

 

ISO 100:

 

ISO 1600:

 

ISO 12800:

 

ISO 25600 (aiming for good exposure):

 

ISO 25600 (worst exposure 2.5-3 steps):

 

ISO 51200:

 

ISO 102400:

 

 

Summary:

I was actually surprised that the loss of quality from 1600 to 12800 is this small. I wouldn't hesitate to go to 3200 or 6400 in the field, even to 12800 if it meant getting the shot. The first 25600 is also better than I had expected it, so if I had to crank up the shutter speed to capture an insect without flash (and without too much details), I might use it. However, if with 25600 the shot is still underexposed, the story changes quite rapidly, as most of the fine detail is lost. It might still be acceptable as documentation, like here if I only wanted to prove the stamp to be the famous Millstatt-stamp, worth millions of Euro, it would do the job (which, sadly is not true, the stamp with a nominal value of 1 Österreichische Schilling, can be bought for 0.45 €). The same is true for 51200 and 102400, documentation can be done, but that's about it, where the latter is almost completely unusable, as was to be expected. Or, if it is already pretty dark and one goes for a spooky photo of an old castle, it might work as well

 

 

Caveats, if you want to apply this to other Canons or other subjects:

1. This camera has got its Bayer filter removed, so on top of increased sensitivity in UV and spatial resolution, any colour effects at higher ISO will simply not be there. I'm sure most of you will have had some coloured high ISO-shot converted to black/white so that it doesn't look quite as awful
   
2. I am everything but an expert in handling photos with a large amount of noise. Using Lightroom, Photoshop, Noise-Ninja, etc. an expert will probably be able to improve the quality considerably. This means, I'm showing about the worst-case.
   
3. The subject is flat, the conditions were perfectly controlled. Of course, this is way to test the performance of the sensor. Out in the field everything changes.
   
4. There will be other caveats which I'm not thinking about right now, as usual
   

[Stefano]

Your dynamic range changes as you change ISO. Here there's a very in-depth video about this:

 

Probably a bit overkill for your needs, but still, it's a nice video.

[StephanN]

Your dynamic range changes as you change ISO. Here there's a very in-depth video about this:

 

Probably a bit overkill for your needs, but still, it's a nice video.

 

Thanks.

 

I know that dyn-range takes a dive with higher ISO, so it's the ever-lasting question of how much a hit a given subject can take, or how much the photog is willing to accept for a given purpose

[StephanN]

I'll perform a field test on insects without flash as soon as possible, hopefully on the weekend

[Andrea B.]

Good testing, Stephan.

I am going to add your specific camera model to the title for accuracy when searching.

 

Question: Where are the 3200 and 6400 tests? I would be interested in seeing anything in the gap between ISO-1600 and ISO-28000 above.

 

Question: You chose 330 nm for the UV-pass? I'm just curious. Why not 360 nm at which it is much easier to shoot outdoors? Also, does that 330 nm UV-pass block the blue?

[StephanN]

I am going to add your specific camera model to the title for accuracy when searching.

 

Question: Where are the 3200 and 6400 tests? I would be interested in seeing anything in the gap between ISO-1600 and ISO-28000 above.

 

Question: You chose 330 nm for the UV-pass? I'm just curious. Why not 360 nm at which it is much easier to shoot outdoors? Also, does that 330 nm UV-pass block the blue?

 

Thanks, Andrea, renaming makes total sense.

 

Here are the 200, 400, 800, 3200 and 6400. I didn't keep any of the ones between 12800 and 25600, because for me 12800 is really the limit of usefulness. If you're interested I can repeat the test and show all ISOs; do you think another subject might be more interesting? I chose the stamp in order to have small details, however, I could easily take some landscape-shot, where details are perhaps less important?

 

ISO 200:

 

ISO 400:

 

ISO 800:

 

ISO 3200:

 

ISO 6400:

 

As to the filter, I was searching for a company providing the b/w-conversion with as much filtering of VIS and IR as possible, and didn't find anybody else then maxmax, and they only offered the 330C. I'm including their transmission curve, which means that IF the sensor went down far enough, a bit of UV-B and UV-C photography might be possible, which I doubt, to be honest. In hindsight, the 330 without C might have been the better choice, because it has a higher transmission and in almost all cases I'm using a second filter in front anyway.

 

[Andrea B.]

Thank you for the additional ISO photos. I had wanted to follow the progression of the noise to see where I might draw the line.

 

That MaxMax chart answers my question about the visible leakage. I have U-330 filter stacks used as "bug vision" because some small amounts of violet, blue and green are passed. So that was behind my question. I either did not know or did not remember that the Xnite line had both a 330 and a 330C. I wonder if the OD blockage number is available for those two filters? A small amount of violet "leak" never hurt a UV photo, but if the blockage is not strong on the blue or green, then I would worry about Visible contamination.

[StephanN]

Thank you for the additional ISO photos. I had wanted to follow the progression of the noise to see where I might draw the line.

 

So, where is your line then

[Andrea B.]

I probably would not go past ISO-800.

 

Added later: OK, maybe I'd go to ISO-1600. It looks quite clean also.

 

But that is just a "quick answer" based on the one series you show.

And my answer is also based on what I typically photograph in UV

which is UV floral signatures.

 

For bees and bees-in-flight, I would toss those limits and start over to see where the balance lies between fast exposures and too much high-ISO noise.

 

I would guess that the actual "jump" is somewhere between ISO-800 and ISO-1600?

Then again a jump after ISO-12800 because ISO-3200, 6400, 12800 look fairly close?

[StephanN]

I probably would not go past ISO-800.

 

Added later: OK, maybe I'd go to ISO-1600. It looks quite clean also.

 

Seems we're on the same page here

 

I could redo the series, starting from ISO 50 and also using the 1/3-steps offered by the camera, so would do 1000, 1250, 2000, etc. as well. Also still want to do the daylight-landscape test in the next week or two, where details are not that important. Especially for the touristic views of cities which will probably never be printed larger than, say 10x15cm, I'd like to know how far I can go - there are times and places where it's not feasible (or even forbidden) to set up tripod & co.