Messages

Yes, really cool. I posted this some months ago in this thread:
https://www.magiclantern.fm/forum/index.php?topic=24461.msg220854#msg220854

Btw pictshare seems very nice. Will try that. 

Definetly better, privacy wise. But not perfect, as we don't own the server and can't know whats going on on their end.

If we are using pictshare now, can someone please clarify how one embeds an image in a post.

Same as before @garry23, using "img":

Code Select Expand

[img]https://pictshare.net/xxx.jpg[/img]
More correct version:

Code Select Expand


[url=https://pictshare.net/xxx.jpg][img width=800]https://pictshare.net/xxx.jpg[/img][/url]


All these are catching my eye not in a good way.

I agree.

Funny you brough this up. For years I had been using Flickr (still do) and although I had noticed majority of the users on this forum were sharing images through ibb.co so decided to get the hang of that since that's what I assumed works best for y'all.

Correct me if I'm wrong though. Ha.

Well, not for me and a1ex:
https://www.magiclantern.fm/forum/index.php?topic=18003.msg216880#msg216880

This scene is not really good for judging the colors. It's actually very difficult to make it look natural as the light sources are mixed and some colors info is missed.

Well, IMO, not everything should look "natural" all the time. For example, the two man behind the musician have purple midtones in it. I prefer to bring those hues to cyan instead, because the combination of yellow and purple are not as pleasing as yellow and cyan...

If I was to judge:
- Your version, @Dmytro_ua, is the most natural posted until now. I would just lift the shadows a bit, to bring more details to his overcoat.
- My version was purposefully exaggerated in the teal/orange. Noise everywhere. Bad separation of midtones in his face. Too much saturation.
- @Bender@arsch version has muddy midtones. I would bring more exposure there and decrese exposure in deep shadows. Colors are good.
- @DeafEyeJedi has some issue with magenta/green. Skin tones look pale.
- @Ilia3101 has a overall green cast.

Of course, each person has it's opinion. This is art, after all.

off-topic: why most of you guys use "ibb.co" to host images? Can't enter the site without javascript, it's quite annoying

Challenge accepted;)

We need someone to judge which one is better now

Nice. I like the fire image.
Indoor scenes has a shift to magenta in highlights, but everything else is good.
Have you tested crop_rec? You always upload in 1080p, so... this might be another step to get better image. Also, uploading to youtube in 10-bit with Rec.2020, as youtube now supports it (there's also this tool for mac os).

You're welcome @70MM13!

It hasn't ever happened @1/50 only when fps override, image fine tuning and 1/48 is set and exposed to specific lighting. So I guess I better stick to 1/50 🏋️‍♂️🎬

Ah ha! Did you do that on your last b&w video? Your last one didn't have this "stuck" motion blur, but this one has. If you're kind, can you please, for the sake of my sanity, give a sample (with motion) between fps-override/shutter-finetunning and one without those (1/50)?

Requires an MLV codec written for Windows.
Same problem with CR3 in Windows. Canon doesn't provide any either.

SageThumbs supports many formats and is open source, although it uses GFL lib, which is proprietary (they could have just used imagemagick or something, don't know why they used GFL). Anyway, Ilia's library could potentially be integrated in one of those thumb viewers for windows...

Regardless of this ISO discussion, the dynamic range is pretty impressive. Even through shadows from clouds and bright sun shines, it still holds all details. Color is good, as always. Think you got the best workflow going.

About the video, nice talk about meditation. I don't practice yet, but I find it hightly interesting for health purposes (some studies link it to higher BDNF levels and less anxiety).
Off-topic, but, would like to recommend an animation called "Mushishi" (in particular episode 12 of season 1, but all of them are good). Think you'll like it.

More exaggerated version
This reminds me of the old rawtherapee forum color grading "competitions".



Configs:

Code Select Expand


<?xml version="1.0" encoding="UTF-8"?>
<mlv_files version="3" mlvapp="1.9">
    <clip file="xxx" relative="../../xxx">
        <exposure>179</exposure>
        <contrast>41</contrast>
        <temperature>3563</temperature>
        <tint>31</tint>
        <clarity>-57</clarity>
        <vibrance>36</vibrance>
        <saturation>10</saturation>
        <ds>0</ds>
        <dr>70</dr>
        <ls>0</ls>
        <lr>50</lr>
        <lightening>0</lightening>
        <gradationCurve>1e-05;0.0001;0.0964187;0.0137741;0.192837;0.0275482;0.336088;0.179063;0.393939;0.30303;0.46281;0.443526;0.548209;0.647383;0.705234;0.84573;0.878788;0.936639;1;1;?1e-05;1e-05;0.0743802;0.00826446;0.206612;0.0798898;0.325069;0.236915;0.440771;0.446281;0.655647;0.663912;1;1;?1e-05;1e-05;0.0826446;0.0550964;0.184573;0.140496;0.297521;0.267218;0.438017;0.426997;0.749311;0.754821;1;1;?1e-05;1e-05;0.170799;0.165289;0.31405;0.327824;0.435262;0.465565;0.53168;0.550964;0.735537;0.782369;1;1;</gradationCurve>
        <hueVsHue>0;-0.00555556;0.077135;-0.0666667;0.181818;-0.405556;0.349862;-0.422222;0.4573;-0.283333;0.581267;-0.444444;0.735537;-0.4999;0.895317;0;1;-0.00555556;</hueVsHue>
        <hueVsSaturation>0;0.0611111;0.0330579;0;0.0798898;-0.00555556;0.157025;0.172222;0.261708;0.105556;0.355372;0;0.498623;0;0.683196;0.283333;0.823691;0;0.986226;0.0166667;1;0.0611111;</hueVsSaturation>
        <hueVsLuminance>0;-0.1;0.0716253;0.00555556;0.859504;0;1;-0.1;</hueVsLuminance>
        <lumaVsSaturation>0;0.0166667;0.407713;0;0.862259;0.45;1;0.288889;</lumaVsSaturation>
        <shadows>22</shadows>
        <highlights>-56</highlights>
        <gradientEnabled>0</gradientEnabled>
        <gradientExposure>0</gradientExposure>
        <gradientContrast>0</gradientContrast>
        <gradientStartX>0</gradientStartX>
        <gradientStartY>0</gradientStartY>
        <gradientLength>1</gradientLength>
        <gradientAngle>0</gradientAngle>
        <sharpen>71</sharpen>
        <sharpenMasking>46</sharpenMasking>
        <chromaBlur>0</chromaBlur>
        <highlightReconstruction>1</highlightReconstruction>
        <camMatrixUsed>2</camMatrixUsed>
        <chromaSeparation>0</chromaSeparation>
        <tonemap>3</tonemap>
        <gamut>9</gamut>
        <gamma>100</gamma>
        <allowCreativeAdjustments>1</allowCreativeAdjustments>
        <denoiserStrength>37</denoiserStrength>
        <denoiserWindow>2</denoiserWindow>
        <rbfDenoiserLuma>41</rbfDenoiserLuma>
        <rbfDenoiserChroma>62</rbfDenoiserChroma>
        <rbfDenoiserRange>16</rbfDenoiserRange>
        <grainStrength>44</grainStrength>
        <rawFixesEnabled>1</rawFixesEnabled>
        <verticalStripes>0</verticalStripes>
        <focusPixels>0</focusPixels>
        <fpiMethod>0</fpiMethod>
        <badPixels>1</badPixels>
        <bpsMethod>0</bpsMethod>
        <bpiMethod>0</bpiMethod>
        <chromaSmooth>1</chromaSmooth>
        <patternNoise>0</patternNoise>
        <deflickerTarget>0</deflickerTarget>
        <dualIsoForced>0</dualIsoForced>
        <dualIso>0</dualIso>
        <dualIsoInterpolation>0</dualIsoInterpolation>
        <dualIsoAliasMap>1</dualIsoAliasMap>
        <dualIsoFrBlending>1</dualIsoFrBlending>
        <dualIsoWhite>21949</dualIsoWhite>
        <dualIsoBlack>8188</dualIsoBlack>
        <darkFrameFileName>No file selected</darkFrameFileName>
        <darkFrameEnabled>0</darkFrameEnabled>
        <rawBlack>2047</rawBlack>
        <rawWhite>5526</rawWhite>
        <tone>0</tone>
        <toningStrength>0</toningStrength>
        <lutEnabled>0</lutEnabled>
        <lutName></lutName>
        <lutStrength>100</lutStrength>
        <filterEnabled>0</filterEnabled>
        <filterIndex>0</filterIndex>
        <filterStrength>100</filterStrength>
        <vignetteStrength>-28</vignetteStrength>
        <vignetteRadius>28</vignetteRadius>
        <vignetteShape>8</vignetteShape>
        <caRed>0</caRed>
        <caBlue>0</caBlue>
        <caDesaturate>99</caDesaturate>
        <caRadius>1</caRadius>
        <stretchFactorX>1</stretchFactorX>
        <stretchFactorY>0.3333</stretchFactorY>
        <upsideDown>0</upsideDown>
        <vidstabEnable>0</vidstabEnable>
        <vidstabStepsize>32</vidstabStepsize>
        <vidstabShakiness>10</vidstabShakiness>
        <vidstabAccuracy>10</vidstabAccuracy>
        <vidstabZoom>0</vidstabZoom>
        <vidstabSmoothing>10</vidstabSmoothing>
        <vidstabTripod>0</vidstabTripod>
        <cutIn>1</cutIn>
        <cutOut>6</cutOut>
        <debayer>5</debayer>
    </clip>
</mlv_files>


By the way, LMMSE got some artifacts:

Nice one. Nailed the setup (well, I like the Contax better, but I love Black Pro Mist). Good colors, no noise, sharp, good contrast.

Yeah, I don't know what caused the flickering. I had FPS Override set to the 23.976 etc. Maybe I'll set 24fps next time? Also had shutter-lock on and fine-tuning although this was the first time I had encountered this. Maybe I should have double-tapped the menu button to refresh? not sure.

Kinda off-topic, but this issue happened *a lot* in this video. This is bugging me, man. It seems like the frames are "stuck". The scene at 35s has this a lot, made me dizzy even. Don't know if that's some issue with my monitor or player. Have anyone counted the number of frames in one second? Maybe it is dropping to something like 19 fps without anyone knowing? Dunno, this doesn't happen a lot in videos from other cameras, only on EOS M.

On the other hand, ISO 111 reduced from Canon's ISO 200, or other low ISOs obtained from iso_regs.mo, are definitely not placebo. These do offer measurable improvements (see first post here), and I've never said otherwise.

Right, so the research was not a waste of time after all. There's room for real image improvements in the ISO research, that's good.
I'll keep myself out of this "war" now

Anyone want to share suggestions or ideas?

I don't have expertise on this, but I think you need a controlled environment, like a closed box with a lamp inside, so that luminance keeps constant. I would also suggest to keep aperture constant, because some lenses don't have accurate aperture rings (particularly old manual lenses). Oh, and I think there was something about manual lenses or too bright lenses (>f/1.0?) that Canon automatically makes the image drop white point (or add +1 digital gain, idk) or something, @a1ex can correct me if I'm wrong.

@Luther: did you increase CA radius? Maybe it even becomes better...

I did, but then it starts to desaturate other tones of red and the shades start to blend together. I've also noticed some color shift while using it. Not too bad, but you can notice even on the example posted above (leafs on the left shift green tones to yellow).

Compiled on Windows 10. Works well. Test taken with 50D (@reddeercity new "4k" build):

Without "CA Desaturate" (click for full-res):


With "CA Desaturate":

MKIII is better on all areas, except camera weight.

Oh yeah, now you're on to something. Great colors, I have no criticism to do, this is as good as it gets. On the 4m30s scene, there's clear color separation, even though the hue range between the red tree (on the right), the yellow tree (on the left) and your skin tones are very close, it still preserved the tones as they are. The shades of blue in your clothes are also well defined, without banding or posterization. This is really difficult to achieve if you don't have a good color manament system as ACES. Well done.
The contrast and sharpness is also spot on, not excessive or too flat. Some scenes have clipped highlights on leafs, but that is normal considering MKIII doesn't have a sensor as good as Alexa LF.

Personally I am only interested in true image refinements. My own tests reveals more than meets the eye and I have since left the iso_reg module for other work.
In your case. Placebo or not, do whatever you like but do not patronize other people's opinions.
On a sidenote. If you're interested in some deeper research around you're settings it could be done. I wouldn't mind to settle exactly what is going on imagewise with these regs. That is we can keep an objective mindset. A1ex provided all tools needed to measure any dynamic range refinements so shouldn't be too hard.

Is it completely "placebo" though? I find it hard to believe all those graphs showing DR improvements are completely fake. A1ex himself said we only have 'rounded' ISO values (100, 200, 400, etc) because of industry standard (or measurements), not because they have the best SNR outcome. Maybe what @70MM13 is doing is not the 'correct' way, but I definetly think there's room for "true image refinements" in the ISO area, it just wasn't explored deep enough.

This capability is not free though - it does come with image quality issues, and working around them turned out to be far from trivial.

That's surprising. I thought it was even simpler, because you don't need to sync the line exposure times...

There's no motor involved, at least not one in the "classical" sense. Your LCD display from your camera doesn't have a motor either - just FYI

Yes, I assumed it worked based on double polarization, which requires filter rotation. This is a different mechanism, it seems to work by one polarizer and one liquid crystal. Wikipedia explains well: " By changing the voltage applied to this crystal filter, it will work as a switching polarising filter, giving different gray scales of the light coming out."

The shutter from the video offers about 4 stops according to its datasheet, and at least 2.5 stops according to that video. Its switching speed is unknown, best estimation: "we'd guess it switches faster than 50ms".

Stacking them and then synchronizing could work... they don't seem to thick. The speed might be limited to the circuit you're using. Might be a nice project to use and learn TinyFPGA, to try to find the actual speed limit.

We can safely exclude #2 and #3. One would only affect motion blur, with no effect on tearing

True. I meant FPS issues.

the other - if present - would have caused frames with tearing, visible when pausing the video.

Yes, and I can't find those. I think "tearing" might not be the correct term. Some articles point to the term "judder". I've tried MPC-HC player instead of MPV (which I normally use) and it still happens. MPV has some options to reduce 'judder' (--tscale), so I'll play with these and see what happens.

This one at 1m55 looks like it has used some stabilisation in post. Looks a little bit warp effect to me, like the stuff you see with digital image stabilisation on phones.

Yes, but doesn't explain the "judder/tearing".

-Rolling shutter and movement from camera. There are differences in rolling shutter with different video modes, especially crop mode has more rolling shutter.

Might be, but @a1ex says this is not the case.

Meanwhile; I'll see if I can back track to an old footage that shows this. If not, I could try to reproduce and then upload certain DNG's for verification/testing purpose.

Would be nice to get actual frames with tearing, but I don't know if that's the case. Might be only some issue with my display refrash rate or my player... Thanks for trying to help @DeafEyeJedi

Anyway, this is probably not related to ML then (even though this only seem to happen on EOS M videos).
I can consider this thread closed.

Nice one. Still has the noise from the previous, though. Are you using some LUT? There's chroma artifacts in blues on some scenes.
I like the shapness, though. The building scenes has some moire, but the sharpness is spot on. Have you tried the "anamorphic mode" in Danne's build?

The images are incredible, love dragonflies.
But unfortunately the post-processing is wrong. Too much sharpen and clarity applied. Also, too much contrast and saturation.

Thanks for clarifying @a1ex.

After the exposure is started by the CPU, the sensor is, indeed, exposed line-by-line, and the time difference between the lines is exactly 1 step (as computed above, and displayed in the FPS override menu).

Correct me if I'm wrong, but expensive cinema cameras (Arri/Red) expose all diodes at once, right? Would it be possible to do the same on known and documented hardware (thinking of Apertus here)?

You can't alter the intra-frame timings, even if you change the settings (such as FPS timers) in the middle of a frame; they only take effect at the next frame. Yes, I've tried that a few times. Canon themselves configure the next frame in advance - as seen e.g. in this graph at 720p60, but can be reproduced at any other frame rate by reducing the HEAD4 timer.

Maybe this is done by another microcontroller running a hard real-time kernel?

The Tessive idea can not be replicated in software, but I've been toying with the idea of modifying one of the software-controlled variable ND filters for this purpose. If the LCD shutter used by that filter is fast enough, it might even work for approximating a global shutter.

Great idea. If the filter works as a double circular polarizer, the speed will only be limited by the motor. Don't know if 24fps is feasible for such tiny motor.

The Aputure DEC Vari-ND appears to have the hardware component I'm looking for (a fast LCD shutter), but I could not find one for the EF or EF-M mount. If you already know where to get a fast high-contrast LCD shutter, ideally already in some kind of lens adapter, please let me know.

Yeah, they seem to only make E-mount and MFT-mount.

Tearing can be caused by incorrect EDMAC settings, or by starting to process an image buffer while it's being written by some other "task" (for example, if the image capture and image compressing transfers "meet" somewhere in the middle of a frame). If it actually happens, show me a way to reproduce the issue. Also see if it happens with any build from the download page.

Strangely enough, I couldn't get frames with tearing. But I can definetly see them while playing. Specific examples: scene at 48s of this video from @ngemu (recorded with EOS-M) and this other from @wib (scene at 1m55s) - Recorded with 5D3 (contradicting what I said before about this only happening in EOS-M). The first is using 23,976fps and the later is 24.000fps. Both downloaded in highest DASH resolution using youtube-dl. MPV player reports my display as 59.978Hz.
Videos where this 'tearing' doesn't happen are this one from @paulheran and most (but not all) of the recordeds with 5D3.

Again, I can't get specific frames. My assumption is that:
1 - Display refresh rate issue (doesn't explain why this doesn't happen on most 5D3 videos)
2 - Shutter speed (don't know the settings each video used)
3 - Some change @Danne did on "EDMAC settings" (I have no technical knowledge in this area, so I can't infer on it)
4 - Placebo or some weird psychovisual artifact (might be, in this case I'm biased - don't think this is the case though)

@Luther: in what application do I find this "advanced tab"? RawTherapee?

Yes:

Is this card defective or am I using the wrong settings?

Had the same issue with Komputerbay 64GB 1000x. Solved by properly formating in a UDMA-7 capable reader (with USB 3.0). You can also try to low-level format inside the camera, might work.
If this doesn't work, I would assume you probably have a broken card.

Did anyone else noticed that? These last weeks some ML users posted their work, some made with EOS M, others with 5D MKIII. One difference (between many) that I perceived was the motion blur. MKIII has a nice motion blur (very cinematic), while EOS M has noticiable artifacts in high motion areas (tearing, to be specific). At first I thought this had something to do with my display refresh rate, but this happens only on EOS M videos.

So I wonder why this happen and if it could be fixed. @g3gg0 already explained sometimes that shutter speed is not precise in canon cameras and that we can't achieve perfect 180 degree like in a cinema camera. But the Timer A and B adjustment of FPS and fine shutter adjustment could solve this tearing? The sensor is exposed line-by-line and not all pixels at once?

Going even further, could this Tessive idea be replicated by exposing pixels in a 'sine' curve instead of square curve? Or this voltage control is too low-level for ML to have access to?

Iso regs for eos m can be reached with adtg_gui.A few at least.
Personally. From my testing with the 5DIII the improvement is minor. Usually what happens when people test(including me) iso signal is pushed and camera still showing original iso. Sort of like exposing to the right ie. iso gain pushed to 400 and then lower gain back to iso 100. So this will of course reduce grain but you loose in dynamic range simply because you're still in iso 400 disguised to iso 100 display and gainwise.
Conclusion: The placebo is a fact.

I still think it was a shame ADTG research stopped. It had potential, if we "brute force" all possible variables.