New camera, first tests and a long story to tell

[Andy Perrin]

Yeah, the usual way is to photograph the same scene with the same settings but with the R72 stacked on there to block the UV and make sure you get a black photo.

[dabateman]

You should try the multi shot noise reduction mode. Apparently takes 4 images and averages them to remove noise.

 

Also according to dpreview it has 31 PDAF points.

[Stefano]

Awesome, that first and second image show you exactly were the PDAF points ate on the sensor in the center.

 

Now you can have some real UV fun.

Yes, noticed it too. I wonder why they come out in this circumstances. Are they more noisy?

 

The camera seems to be quite sensitive in general. I even tried running my 340 nm LED at 100 mA (low power since it’s without heatsink) and I can literally light up my bedroom at 10-20 minutes exposures. I noticed I can’t set an in-camera white balance with it though. Setting a sunny white balance, a 365 nm torch appears reddish magenta, almost pure red, and that LED is just red. I should try shooting RAW to see it’s true RAW color, deeper UV should appear green in RAW, at least in some cameras.

[dabateman]

Yes should be. They are not full pixels as older system, not the newest dual pixel version. Image Resource says this:

"Canon EOS M offers on-chip phase detection autofocus from the main image sensor, or Hybrid CMOS AF in Canon parlance. The system, used both for still and movie capture, combines phase detection and contrast detection when the subject is near the center of the image frame, and uses only contrast detection when the subject strays outside this area. When phase detection is used, it's only to determine the focus direction and get focus in the ballpark; focus is always fine-tuned with contrast detection towards the end of AF operation.The Canon EOS M's autofocus system operates in one of three modes: Face Detection & Tracking AF, Multi-point AF, or Single-point AF. A total of 31 autofocus points are available."

[colinbm]

Good to see you underway again Stefano

[bvf]

 

I tried some very long exposures in BULB mode:

 

 

How did you do your long exposures, Stefano?

 

IIRC, the EOS M doesn't work with the Canon rermote control (which allows you to use one-click-open, second-click-close when the camera shutter is set to B(ulb) - like the T(ime) setting back in the day).

 

I made a device out of bent metal and screws which physicaly held the shutter down. I've still got it - if it's of any use to you.

[Stefano]

I enabled touch shutter and touched the display twice.

[bvf]

I enabled touch shutter and touched the display twice.

 

Well, there's something I didn't know!

[Stefano]

Just to give it a try, I made a tri-color IR image (as described here) and attempted a bi-color UV one, with red at 365 nm and cyan (green+blue) at 340 nm, using my LEDs. I used red+cyan to have white when stacked.

 

IR:

 

UV:

 

In the IR, the 940 nm exposure looked "softer", but this is probably because of noise and not because of focus shift. In UV, instead, I think the 340 nm exposure is a bit out of focus (I focused at 365 nm).

 

I used pretty high ISOs for the images (3200 for the IR and 6400 for the UV image). Also note the objects are not all the same, the camera is not in the same place etc. This was just a test. I would like to use the wavelengths Bernard used in UV (I think he used 380, 340 and 320 nm, should check), but I could also try 385, 365 and 340.

 

I used my usual UV stack (ZWB2 (2 mm) + Chinese BG39 (2 mm)) for UV and Hoya R72 for IR, to be more sure the signal was clean.

[dabateman]

My goodness the uv transformed that rubrics cube into some unknown thing. I haven't seen that before.

 

385(blueish), 365(yellowish), and 340(greenish) should work well.

You should try the multi shot mode to see if it cuts down on the noise, or is it limited. I don't know.

[Stefano]

Yes the Rubik's cube was added later (I did the UV first). Also I now noticed the images are a bit underexposed. For the IR I just put the camera in auto exposure, so that I didn't have to figure out the appropriate values. I will change the exposure compensation.

 

I think I could try with a lower ISO to reduce the noise. Maybe I will also try the multishot to see if it improves things.

[bvf]

Stefano - my filters were 380BP20, 345BP25, 315BP20 - although the peak transmission of 315BP20 is about 320nm.

 

It will be interesting to see how how yur future tri-colour UV compares with this bi-colour. In my tri-colour efforts plastic comes out red-brown. I also did a bottle of alcohol (IPA) and the translucent bottle was white but the cap was red-brown. Your bottle cap looks more black, but you have the red-brown in the plastic lens (I think it is) and the bottle of roll-on deodeorant.The CFL light is interesting - I haven't seen that in tri-colour UV before.

[bvf]

 

Also I now noticed the images are a bit underexposed. For the IR I just put the camera in auto exposure, so that I didn't have to figure out the appropriate values. I will change the exposure compensation.

 

 

Yes, I've found on the A6000 that both UV and IR require +1 to 2 stops of exposure compensation when using autoexposure. I assume you're working with RAW, so you could probably boost the exposure in post processing without too much problem.

[Stefano]

I am not working with RAW yet, I really should start doing it.

 

The lens is plastic. It appears purple/lavender under sunlight (in UV). At 340 nm it's completely black.

 

The CFL surprised me too. I don't know if it is the glass or the inner coating that absorbs deeper UV, I know these bulbs are (at least in theory) designed to block UV, but in practice the one I have emits a significant amount of UV, including the 365 nm I-line.

 

I am now curious to try the cube in UV. I may have reds with this technique. Some squares will surely be all-black.

 

One day I would really like to reach at least 320 nm, if my lens allows it. If its spectrum is this https://www.ultravioletphotography.com/content/index.php/topic/3412-relative-transmission-measurement-of-two-35mm-lenses-prinz-galaxy-and-soligor then it will be doable. If I only had a spectrometer...

[Andy Perrin]

Yeah, work with RAW, I never touch JPGs with a ten foot pole until the last step (exporting to UVP!). You really need 16 bits or you lose too much information when you manipulate the images. Also, you are correct that the images are very underexposed. It's actually better to deliberately OVER expose, then use the RAW to bring the image back into the right range. This is called "expose to the right" or ETTR.

[Andy Perrin]

My goodness the uv transformed that rubrics cube into some unknown thing. I haven't seen that before.

It's because he rotated the filters!

[dabateman]

 

It's because he rotated the filters!

 

Oh yes that makes perfect sense now. That London city hall shape was hiding inside the tesseract. No doubt the Doctor was using it to hide someone.

[bvf]

 

One day I would really like to reach at least 320 nm, if my lens allows it. If its spectrum is this https://www.ultravio...axy-and-soligor then it will be doable. If I only had a spectrometer...

 

 

I think your lens will be OK. The problem will be finding a filter, now that Omega has stopped providing them.

[bvf]

 

It's actually better to deliberately OVER expose, then use the RAW to bring the image back into the right range. This is called "expose to the right" or ETTR.

 

 

What is the background to that, Andy?

 

Most of my stuff these days is UV macro, using flash. And I'm plagued by burnt-out highlight spots where the flash is reflecting off surface spots. so I tend to underexpose to try to reduce this effect and then add a bit of positive exposure compensation in the RAW. But I still find that at the final TIFF stage (i.e. before I go to JPEG) I still need to suppress highlights and boost shadows.

[ulf]

I think he is referring to the exposure level reported by the cameras.

They often have a safety margin, showing a higher exposure than the sensor's full range, and are in some cases only using the channel that is most often overexposed in typical VIS situations in un-modified cameras.

Over exposing of any channel is a loss of real data and colour information.

Some overexposure can be recovered by guesswork in the processing of the RAW-information, but is still a guesswork.

 

By working with RAW-files and doing alternative exposures you can find out how much over exposure is tolerable

 

Different types of filter stacks has been more or less easy to handle with this.

Expose to the right means not to over expose, but expose the channel with most as much as possible without overexposing it.

What the camera say and the real data often differs.

 

If I remember correctly RawDigger is a good tool for this.

https://www.rawdigger.com/

 

Many raw processing programs have an ability to show the channel histograms.

I mainly use FastRawViewer for this to quickly evaluate my RAW-files.

https://www.fastrawv...t-and-features#

[Stefano]

Just tried shooting in RAW+JPG, and the RAW file has a .CR2 extension. It retained the white balance I put, so I don't have "raw" colors. At least, this is how it opens up in the Windows photo viewer.

[bvf]

Just tried shooting in RAW+JPG, and the RAW file has a .CR2 extension. It retained the white balance I put, so I don't have "raw" colors. At least, this is how it opens up in the Windows photo viewer.

 

Yes, if you view in Photos it will apply the camera white balance. You'll need to use RAW-capable software to handle the .CR2 properly. I'm sure you'll get bombarded with advice - in my case I use RawTherapee (free). (I think someone said they don't use it because it handles only 8-bit TIFF, but actually the current version can handle 32-bit TIFF. As well as many RAW formats and JPEG, of course.)

[ulf]

Normally the raw-file gets a pdf-preview appended in the file and also the camera's WB-setting.

I think you still are looking at a pdf.

 

As I shoot with many different filters I use that to visually identify the filters by looking at the preview and keeping the camera's WB-setting.

The disadvantage is that I cannot see a WB-ed image on the cameras screen, but that is not important for my way of working.

[Stefano]

Thanks. I installed RawTherapee and opened the image, but I still see the white balanced colors. How can I see the "true" colors? I did a quick search online and didn't find how to do it.

[dabateman]

Thanks. I installed RawTherapee and opened the image, but I still see the white balanced colors. How can I see the "true" colors? I did a quick search online and didn't find how to do it.

 

There aren't any. In UV the white balance settings matters. White balance applies signal to each of the channels. If your white balance is way off you can actually clip a channel without knowing it since UV will have strong sensitivity in only one channel which changes the deeper into UV you go. If the WB is off it will be hard to recover the image in software.

If you don't know the white balance and don't want to clip, I have shot in camera monochrome mode to get the exposure correct, but its still tricky to tune in the colors I want later.

Best to try and set a correct WB while you are photographing.

 

If you want to inspect the raw data channels, than raw digger is easiest. But expensive. ImageJ will get there too with the dcraw plugin and and update dcraw file copied in to the plugin folder.

http://ij-plugins.sourceforge.net/plugins/dcraw/

 

There is also Histogrammer which may not run on your computer, its very old software.

http://www.guillermoluijk.com/software/histogrammar/index.htm

 

For editing just have fun in RawTherapee or Art (a fork of RawTherapee). They give you a lot of control over your images.

 

I also like photoscape x. But its not the best for WB recovery. Just has a lot of cool things there designed by photographers.