Dual ISO - massive dynamic range improvement (dual_iso.mo)

[wavydoom]

Disabled all modules on my 6d (Made in Japan, it says on the bottom).
Only activated dual iso module.

Got the 'ISOless PH err(3)' error both in video mode and photo mode...
So it doesn't work in any mode.

Will try some other builds of magic lantern.

1.1.3  with   magiclantern-Nightly.2014Mar28.6D113     

dual iso works fine.

[Walter Schulz]

Problem exists (at least) for 6D, 650D, 700D. According to a1ex it should be fixed by solving the common issue: https://www.magiclantern.fm/forum/index.php?topic=25976.msg231177#msg231177

[Levas]

I think something changed from 1.1.3 to 1.1.6 on 6d.
Anyway, the  'ISOless PH err(3)' for 6d is fixed in the builds I uploaded the last couple of years, although those aren't available on the experimental downloadpage.

In this post there's a linkt to the latest build for 6D I uploaded, should work for dual iso:
https://www.magiclantern.fm/forum/index.php?topic=25782.msg229094#msg229094

But be aware that this is a experimental build, this build also contains the SD_UHS speed hack module and has a special crop_rec module with more resolution options.
To know how it all works, you might want to read the whole topic, it's only 6 pages, so it's not that much. But a lot of info in there.
https://www.magiclantern.fm/forum/index.php?topic=25782.msg223031#msg223031

Comes to mind that the dual iso module in these builds should be compatible with the 4k build on the experimental downloadpage, so if you don't want to use all the stuff in the new build, you could copy the 'dual_iso.mo' in the modules folder of my build and overwrite it to the 4k experimental build you are using from the downloadpage.

[dream951]

Hi guys!
Is it possible in any build to take silent pictures from liveview with dualiso with an intervalometer.
This does not work for me, it shoots out only a raw frame without processing by dualiso.
It would be very cool for timelapses.
Thanks

[theBilalFakhouri]

Hi @dream951

Silent pictures with Dual ISO already works:

1-If you set "Silent Mode" to "Simple", you need to enable RAW video in this case to make Dual ISO works with silent pictures (this only works in movie mode, unless you want it in Photo mode for slower shutter speeds?)
(*Edit: it seems this doesn't work well with intervalometer, I got "Out of Memory" with RAW video enabled).

*Workaround: make Dual ISO works when taking "Simple" silent mode pictures, also when "Silent" mode is activated in movie mode without RAW video enabled (coding tasks).

2- If you are using "Full-res" silent pictures, it works out of the box, no RAW video required.

"Simple" silent mode has LiveView RAW resolution which is about ~1920x1280 or less (in none crop mode, depending on your camera) with Binning/Skipping, "Full-res" silent mode uses full pixels on sensor without Binning/Skipping, which one are you looking for?

[Milk and Coffee]

Does MLV store RAW data in a linear gamma?

I’m reading that cameras that store RAW sensor data linearly, the dynamic range is limited to the bit depth. Whereas cameras that store RAW sensor data logarithmically, the dynamic range is not limited by the bit depth.

https://www.avsforum.com/threads/fake-raw-braw-proresraw-z-raw-whats-going-on.3157980/

Example: In cameras that store sensor data linearly, capturing a 10bit signal, limits the available dynamic range to be captured to 10 stops.

[ilia3101]

Technical doc
- dual_iso.pdf (in-depth description of how it works)
- See also: ADTG and CMOS discussion

The technical doc link is down. However it can still be accessed here if anyone needs it: https://web.archive.org/web/20130821183804/http://acoutts.com/a1ex/dual_iso.pdf




@Milk The developers would have done it if there was an easy way to make the camera do it! 10 bit log would be almost indistinguishable from 14 bit.

[2blackbar]

Milk and Coffee 10,12 and 14 bits in raw on my canon M or 5d2 are hard to tell apart, 10 bits have noisey blacks so i use it when i have no choice and footage is too high res for continuous recording
Dynamic range is identical on all 3 bitdepths, i tested it, there wasnt more stuff and even if there was its just junknoise ,you can test it easily too.I use 12bits.

[dragonsfire1981]

I tested this mode for the first time today, I seem to have got a few strange corrupted frames on 1080p mode.

Is this less likely to happen if I use a crop mode or is it suspectable to it regardless of mode?

[yuvex]

Astro modified 550D owner, and new Magic Lantern user here.

I have the Dual ISO module successfully installed and running, and have taken a few one shot daylight HDR images.

Fast forward to using this in my deep sky tracked and stacked astrophotography workflow, and I am struggling with image calibration.

Basically, my flats are overcorrecting and not removing any dust bunnies. There also appears to be weird multi coloured colour casts.

I am at 100/1600, Dual ISO-wise for what it's worth.

And yes, I have tried calibrating the files with and without crwhdr and every combination in between, without success. Without flats, the overcorrection is gone, but then you run into all the other incomplete calibration issues this brings.

I know this is a long shot, as I'm not sure anyone without DSO astrophotography knowledge has any clue what I'm talking about, but there are no Dual ISO users in my astro circle I can call upon.

Thanks

[Walter Schulz]

I don't think you will find any information on that. Dual-ISO and astrophotography is IMHO not a combination anyone wants to use.

[vastunghia]

Yuvex,

For sure not a DSO photography expert here, but I know what you are talking about.

May I ask you why one should be using dual ISO for astrophotography? You are losing half vertical resolution in deep shadows (that is, the part you are most interested in) and deep highlights.

Alternatively, you can ‘just’ double your overall exposure time and take (if you really need to!) a set of shots at ISO 100 after taking the one at 1600. Without losing resolution.

Though I’m wondering what kind of DSO subject would need a 4-stop exposure bracket. But once again, I’m no expert here. Would be nice to learn something interesting new on astrophotography.

The way I see this, dual ISO makes extremely sense when you have a moving subject, that makes it impossible to apply standard HDR bracketing techniques. DSO’s do not fall in this category

But I may be missing something.

S

Ps: the way dual ISO is debayered is not trivial. I guess this could also make calibration impossible. Anyway, what kind of calibration workflow are you adopting? Are you going through flat darks as well?

[yuvex]

Hi, thanks for responding.

Under normal circumstances, my 550D has a peak DR of 10.8 stops at ISO 100 (I normally image at ISO 400). By current standards, this is fairly mediocre.

I'm aware by stacking I am increasing my SNR, and also my DR, but why not try to maximize it further with Dual-ISO too? I am not green when it comes to astrophotography, so look to any avenue to better my improve my workflow.

Resolution is not usually an issue, as more often than not I BIN my data to increase SNR (my data is often oversampled out of the box, so I am not really losing anything but useless noise).

I have successfully taken a Dual-ISO widefield astro panning time lapse, but that needed no calibration frames. That dealt with bright foreground artificial lighting and the night sky very nicely. The problems lie when you apply calibration frames for deep sky.

Calibration wise, I use lights, temperature and time matched darks, flats (via an EL panel) and bias frames. All pre processed in Siril.

I only employ dark flats over bias for my dedicated cooled monochrome astro camera, as bias are sufficient for an old DSLR like the Canon.

[yuvex]

I don't think you will find any information on that. Dual-ISO and astrophotography is IMHO not a combination anyone wants to use.

It seems that way.

Maybe I'll just stick with terrestrial and uncalibrated time lapse.

Cheers

[stephanW91]

Why did CR2HDR export a DNG with a resolution of 384x252 while still being more than 37mb? what can i change to improve this?

[Walter Schulz]

Open DNG in RAWtherapee first. I think you will see a difference.

[magellan]

I was looking on the PhotosToPhotos.com website recently to see if there was information there on the dynamic range of the 5D Mark III with dual iso.

I was surprised that the 5D Mark III wasn't listed, (though there was info for the Mark II), so I contacted the website owner and offered to assist him with gathering data for the Mark III, which he has now posted to the website. Should you be interested, here's the link comparing the standard Mark III with ML dual-iso base settings of 100 and 400 ISO : https://www.photonstophotos.net/Charts/PDR_MagicLantern.htm#Canon%20EOS%205D%20Mark%20III,Canon%20EOS%205D%20Mark%20III_16(ML),Canon%20EOS%205D%20Mark%20III_16(ML400).  The results suggest a dynamic range improvement of approx 1 bit over the 5D Mark II.

Its not possible to compare the 5D Mark III with the much more recent Canon R5 on that chart, but reading the separate chart for the R5 here : https://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%20R5 suggests that the Mark III does well for dynamic range, albeit with a lower resolution.

[names_are_hard]

Cool, thanks for getting another cam added to their list!  Are there other missing cams that can do dual ISO?

[Danne]

On a sidenote. R5/R6 behaves practically like dualiso already included in the sensor output meaning shadow handling on these cameras is not even comparable anymore to digic V sensors.
Do I still pick up my eos m though. Hell yes.

[Walter Schulz]

Are there other missing cams that can do dual ISO?

A lot:

ML supported cams not listed at all:
1200D
1300D
500D
650D
EOS M2


Missing Dual-ISO data:
100D
1100D
50D
60D
700D
70D
EOS M

Missing data for EOS M, M2, 650D, (1200D/1300D) is no big deal for those in the know. They share 600D's sensor AFAIK. Same for 100D, 700D.