Homemade method of simulating FILM with Leica M9 - Testers & feedback welcome!

[camalogica]

Advertisement (gone after registration)

On 11/25/2019 at 9:01 PM, JonPB said:

My only programming-related request is the ability to forgo grain simulation.

Hi again,

I've added the -nograin option to the command line to generate TIFF files without grain.

It's possible to download new version (1.03) from the following link:

https://mega.nz/#!ldlBBaCD!bp51YxHzXoormGjQ1qCPUicXcrIW1GTYGMEe_UB1-ic

Kind regards,

Tomás

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[JonPB]

I like it. I ran a few images through it for black & white processing and found the output to be rather nice with ``-autodevelop'' and, in Capture One, adjust levels (black point, white point, and set "exposure compensation" with middle point), then fine tune with a simple curve adjustment.

I recommend this particularly to Monochrom shooters. I ran an M9M file through it and while a few highlight details were lost in an already bright file, the image immediately lost that digital flatness that I sometimes struggled with.

There are definitely things going on here that I can't quickly replicate in Capture One. There's a degree of contrast that I often try to attain in post that comes out immediately with the Tri-X sim. No sharpening is applied; detail extinction is perfectly smooth to start with and easy enough to adjust with a sharpening tool. Sometimes noise is emphasized but in a rather flimic way that looks appropriate for a pushed image, not distracting.

Superia looks good. I'm not as experienced with color as B&W, but somehow my M9 DNGs are now looking decidedly ... digital. There are a few interpolation issues at the pixel level but nothing I find troublesome.

I want to run some tests to see how this program responds to differences in exposure and ISO. Which is to say, I feel like more effort in learning how this program changes images will be well spent. If I find anything interesting, I'll share some examples.

Two more thoughts for possible improvements: retain EXIF information (possibly with credit for the software) and enable LZW compression. I'd also be interested to see if any other films strike you, Tomás, as being worth simulating.

Thank you!

[camalogica]

9 hours ago, JonPB said:

I like it. I ran a few images through it for black & white processing and found the output to be rather nice with ``-autodevelop'' and, in Capture One, adjust levels (black point, white point, and set "exposure compensation" with middle point), then fine tune with a simple curve adjustment.

I recommend this particularly to Monochrom shooters. I ran an M9M file through it and while a few highlight details were lost in an already bright file, the image immediately lost that digital flatness that I sometimes struggled with.

There are definitely things going on here that I can't quickly replicate in Capture One. There's a degree of contrast that I often try to attain in post that comes out immediately with the Tri-X sim. No sharpening is applied; detail extinction is perfectly smooth to start with and easy enough to adjust with a sharpening tool. Sometimes noise is emphasized but in a rather flimic way that looks appropriate for a pushed image, not distracting.

Superia looks good. I'm not as experienced with color as B&W, but somehow my M9 DNGs are now looking decidedly ... digital. There are a few interpolation issues at the pixel level but nothing I find troublesome.

I want to run some tests to see how this program responds to differences in exposure and ISO. Which is to say, I feel like more effort in learning how this program changes images will be well spent. If I find anything interesting, I'll share some examples.

Two more thoughts for possible improvements: retain EXIF information (possibly with credit for the software) and enable LZW compression. I'd also be interested to see if any other films strike you, Tomás, as being worth simulating.

Thank you!

Thank you very much for your comments!!

Only for info, even if it supports all input DNG ISOs, in theory, it should be used with DNGs having the same ISO as the result film.

For example, for TriX 400, the program is done to match Leica M9 ISO 400 DNGs. Any other ISO will work, but should differ a bit more from reality.

Tomás

[camalogica]

BTW: The autodevelop option does the following:

- Apply Gamma 2.2

- Invert channels

- Auto-levels, similar way that GIMP does it.

So... nothing fancy! 

Tomás

[JonPB]

ISO series with TriX 400 simulation. This is a Capture One 9 screen shot showing ISO 200, 400, 800, and 1600 shot thumbnails (a) SOOC JPEG (with medium-high contrast), (b) with the DNG -> m9tofilm -> TIFF -> inverted, and (c) as 'b' plus auto-levels. Shutter speeds were set by the camera and formed an expected progression of 1/45, 1/90, 1/180, and 1/360.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

(The image was compressed in macOS Preview to fall below the 1/2 megabyte forum limitation.)

Shooting at a lower ISO does not seem to hurt anything, although there is less shadow detail (hard to see here). Shooting at a higher ISO definitely compresses the results into a narrower slice, and while a lot of information can be retained there is still a lot of loss here. My takeaway: shooting at the ISO expected by the program is worthwhile.

[JonPB]

Exposure series (+/- 3 stops) with TriX 400 simulation at camera ISO 400. This is a Capture One 9 screen shot showing (a) SOOC JPEG and (b) DNG -> m9tofilm -> TIFF -> inverted thumbnails. Each pair represents a full-stop interval with the "green" selection at the camera's auto-metered suggestion of 1/180.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

The program brightens each image significantly; by my eye here, it looks to be about two stops worth of brightening. This makes sense as the shadows are lifted to better balance shadow detail with highlight detail, or perhaps to offer more highlight information than shadow (like film negatives). My takeaway: shooting with the intent of using m9tofilm for initial development means that I'm working at a camera ISO of 400 plus an exposure compensation of +2, which is essentially shooting at ISO 1600. I'll give that a try next time I'm out, which seems worth trying if the quality is there.

[JonPB]

Advertisement (gone after registration)

Detail comparison of 1/180s DNG -> m9tofilm -> TIFF on left with 1/45s SOOC JPEG on right; at 100% in Capture One 9, screenshot compressed (but not resized) in macOS Preview. The m9tofilm image has been inverted, auto-leveled, and modestly sharpened (see panel in image).

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Two stops of brightening looks about right to me here in the midtones. The 1/45s shot -- two stops faster than the M9 meter suggested -- has blown out the sky entirely. What strikes me as more important is how gently the m9tofilm image dissolves toward 0% contrast at pixel resolution (which I suspect is due to its demosaicking process), which is similar to how film dissolves into mere graininess at maximum resolution, and which contrasts against the relatively unpleasant digital acuity of the SOOC image. (I can also approach that level of acuity with additional, optional sharpening.) Yes, the m9tofilm version has much lower contrast, but that is easy to add with a few sliders. A gentle S curve is all that this photo requires to achieve a nice punchy B&W image.

I did attempt to mimic the m9tofilm result using the same DNG file but was unable to achieve satisfactory results after about 15 minutes. The m9tofilm image is pleasant, easy to create, and easy to work with. My next step will be to aim for this workflow on my next walkabout shooting session. The real test for any tool is how well it serves actual photography, though for now I'm optimistic that this will be quite nice. Tomás' program offers a lot of the subtlety of film with a functional sensitivity of ISO 1600 and final resolution that strikes me as perhaps comparing favorably with 645 or perhaps even 6x9 film. I remain impressed.

 

Edit: D'oh. Looks like the forum has scaled my images to a more reasonable size. If there is interest in full resolution copies of the above (or other shots), I can post them on another host.

Edited December 8, 2019 by JonPB

[oldwino]

This is very cool. I hope we Mac Catalina users will get a version soon!

[camalogica]

14 hours ago, JonPB said:

ISO series with TriX 400 simulation. This is a Capture One 9 screen shot showing ISO 200, 400, 800, and 1600 shot thumbnails (a) SOOC JPEG (with medium-high contrast), (b) with the DNG -> m9tofilm -> TIFF -> inverted, and (c) as 'b' plus auto-levels. Shutter speeds were set by the camera and formed an expected progression of 1/45, 1/90, 1/180, and 1/360.

Shooting at a lower ISO does not seem to hurt anything, although there is less shadow detail (hard to see here). Shooting at a higher ISO definitely compresses the results into a narrower slice, and while a lot of information can be retained there is still a lot of loss here. My takeaway: shooting at the ISO expected by the program is worthwhile.

Hi, I like all your testings and your conclusions about the program

As said before, all the chemical simulation is done for the native film ISO. So, for TriX 400, ISO 400 should be always used to get the closest thing to the reality. (Since all the simulation calculations done for film behaviour were precalculated using this params).
It's true that it's possible to push the image (+/-3 stops in camera) and then, at developing time, push the image the same number of stops. Only note that if using -autodevelop, this won't be optimal.

My workflow for TriX400 is the following:
1. Shoot the DNG at ISO 400 pushed +/- 3 stops (or not).
2. Develop using: m9tofilm.exe -i "E:\DATA\LeicaM9\dng\160\02_CABANAL" -o "F:\DEV\output" -f trix400 -frame false
3. Develop using Silverfast HDR (https://www.silverfast.com/show/silverfast-hdr/en.html)
4. Final adjustments using Lightroom

Anyways, thank you very much for testing with so much detail!! Im really glad you like the output of the program 🙂

[camalogica]

9 hours ago, oldwino said:

This is very cool. I hope we Mac Catalina users will get a version soon!

Unluckly I'm quite busy and it should take me some more time to port it.

In the meantime it's possible to use it with "wine".

Tomás

[willeica]

Look good. Fuji has film effects filters in Lightroom and there are also filters for all cameras in Nik Color and Silver Effex. The existing effects are quite good, particularly the Fuji ones, but they could all be improved upon. I will look at what you have done and compare. The best way of getting film effects is, of course, to use film. It is still around and getting increasingly popular, particularly with young photographers.

William

[oldwino]

3 hours ago, camalogica said:

Unluckly I'm quite busy and it should take me some more time to port it.

In the meantime it's possible to use it with "wine".

Tomás

Thank you. Being busy is mostly always a good thing. Take your time - I still shoot plenty of film...

My understanding is that Wine won’t work with the latest OS X version, Catalina, as only 64 bit is supported. Unless Wine has been updated in the meantime.

[camalogica]

2 hours ago, oldwino said:

Thank you. Being busy is mostly always a good thing. Take your time - I still shoot plenty of film...

My understanding is that Wine won’t work with the latest OS X version, Catalina, as only 64 bit is supported. Unless Wine has been updated in the meantime.

Unluckly you are right. Apple killed 32bit support, so... wine is dead in osx catalina  

 

[JonPB]

7 hours ago, camalogica said:

...all the chemical simulation is done for the native film ISO. So, for TriX 400, ISO 400 should be always used to get the closest thing to the reality.

I haven't shot black and white film, much less Tri-X, recently enough to criticize this simulation -- it is perfect for all I know. However, if I use the camera's meter to create a middle gray, the result is a little less than two stops brighter than middle gray after running it through m9tofilm. I'm used to thinking in terms of gray input to gray output, so pulling two stops of exposure (which is -2, not +2 as I said earlier) fits into my shooting habits. That brighter gray may well be a more accurate representation of how Tri-X would perform with the same lighting and exposure, and I might benefit from rethinking my exposure calculation process if I use m9tofilm frequently. For now, though, I'm just looking forward to seeing how this works with real photos, not just test shots.

If anyone is interested in creating quick and easy previews while retaining the original TIFF, I've been using ImageMagick:

$ convert file.tif -negate -modulate 100,1 -quality 100 -contrast-stretch 1 output.jpg

(Modulate changes brightness, saturation, and hue levels, which inverts with 'negate,' so 100% brightness means no change there and 0% saturation gets us to B&W. Unfortunately, Capture One doesn't play nicely with any gray colorspace yet ImageMagick auto-converts desaturated images, so I use 1% saturation here. Setting quality to 100 is likely unnecessary as the default is 92, but that's how I roll. Contrast-stretch is an auto-levels feature that sets the given number of pixels to absolute black and white.)

I'd offer to help you, Tomás, port the program to be compatible with POSIX-ish platforms, but I'm just getting up to speed with C and so having me muck about with your C++ code and build system would probably be more hassle than it is worth. :-) Maybe I'll look at what it would take to put together a QEMU image that would run a minimal Linux/Wine stack.

[camalogica]

51 minutes ago, JonPB said:

I'm used to thinking in terms of gray input to gray output, so pulling two stops of exposure (which is -2, not +2 as I said earlier) fits into my shooting habits. That brighter gray may well be a more accurate representation of how Tri-X would perform with the same lighting and exposure, and I might benefit from rethinking my exposure calculation process if I use m9tofilm frequently. For now, though, I'm just looking forward to seeing how this works with real photos, not just test shots.

One thing to note is that output, if no -autodevelop is used or -gamma is set, by default, to GAMMA 1.0

This means that the negative file will be really dark, so, when inverting the negative, it will be really bright. Could it be this the reason of that two extra stops??

Maybe worth trying -gamma 2.2

Just my two cents!!

Tomás

[JonPB]

Thanks for trying to help me figure this out!

I've been using "wine ./m9tofilm.exe -i ./input/ -o ./output/ -f trix400 -frame false -gamma 2.2 -nograin" to process the files. The ``inspect'' tool of ImageMagick reports a gamma value of 0.454545 for both the TriX_400.tif files and my ``convert'' inverted files, although I'm not sure that's testing what I want it to test. I do not see any changes if I specify gamma values in the convert process either, so I think something's not quite right in how I'm using ImageMagick. I suppose the definitive test would be to run a print (where my software applies a gamma of 2.2), but I'm not quite there yet.

If I use ``-gamma 1.0'' or omit the -gamma argument with m9tofilm, I definitely see a darker negative than the -gamma 2.2 files I've been talking about. With ``-autodevelop'' and no other parameters (except input/output folders), the result looks the same (plus border and grain) as my results with -gamma 2.2. So I think the program is behaving as expected, but perhaps something is happening with Wine.

If you take a DNG & JPEG shot and run the DNG through m9tofilm with Tri-X simulation, does it have the same midtone brightness as the JPEG version?

[camalogica]

11 hours ago, JonPB said:

"wine ./m9tofilm.exe -i ./input/ -o ./output/ -f trix400 -frame false -gamma 2.2 -nograin"

Your command is perfect 🙂

Output file will be using gamma of 2.2 (as it should), so if you see the image brighter maybe there is a problem in (my) camera calibration. I should recheck.

Thanks for the testing!! 🙂

[JonPB]

The following is a virtual contact sheet from some walking around shots (nothing really successful), shot at ISO 400 with the M9 set to -2 exposure compensation (but me handling metering like I otherwise would, sometimes on A and sometimes tinkering), developed with m9tofilm (-f trix400 -frame false -gamma 2.2 -nograin), and auto-processed with ImageMagick (convert ... -contrast-stretch 1% -negate -modulate 100,1 ...), then previewed in Capture One 9 for this screenshot. Mostly shot through my 21/2.8 Asph.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Overall, I have to say I'm very pleased. When I look at the shots, I'm thinking first and foremost about what I did, not about what the technology has done, and that's as it should be. I haven't closely compared these shots to other alternatives -- Leica's in-camera JPEGs, Capture One auto settings, or the way I like setting my Fuji OOC JPEGs -- but this quality certainly merits further inspection.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[camalogica]

6 hours ago, JonPB said:

Overall, I have to say I'm very pleased. When I look at the shots, I'm thinking first and foremost about what I did, not about what the technology has done, and that's as it should be. I haven't closely compared these shots to other alternatives -- Leica's in-camera JPEGs, Capture One auto settings, or the way I like setting my Fuji OOC JPEGs -- but this quality certainly merits further inspection.

🙂 Thank you very much!!! I like the way your shoots look!!!!

 

[proenca]

quite interesting topic !

will later on install wine and see how this goes