Messages

Something along these lines: http://www.guillermoluijk.com/article/nonoise/index_en.htm

off-topic, but I was trying to find a link to download the Histogrammar utility described ... (for Win-7-64 in my case). Available?

Or can the equivalent viewing of 16-bit histograms be done with RawDigger?

The amount of resolution loss is directly related to the distance between ISOs.

DR increase is directly related to the distance between ISOs.

Need more DR, increase separation between ISOs, also suffer increased resolution loss, etc.

Is there a way to tell what the "effective resolution" is when using DUAL_ISO?

With Adobe Bridge, the metadata with my 5d2 continues to show nearly 5616x3744 resolution in the CR2 and the generated DNG (from cr2hdr.exe), regardless of whether I use 100/200, 100/400, 100/800, or 100/1600.

Is that "as expected"?

I also tried with Canon's DPP, and with RawDigger to see if they reported a significantly different resolution. No difference.

BTW, I noticed with RawDigger, that the .CR2 files had max channel values of about 14,740 (same as when not using DUAL_ISO). However, the generated .DNG from 100/1600 had max channel values of about 57,160. I suppose that is the blend of 100/1600? Ingenious / magic. Well done.

a1ex: Thanks!

I'll give the exiftool command lines you provided a try.

My only use of ExifTool before was filling in missing values for f-stop and focal-length for .DNG's taken with my Samyang 14mm (great lens imo) and Samyang 8mm fisheye (very good lens).

TMI?
I always used ExifToolGui and never tackled the learning curve of ExifTool itself. I am concerned now that ExifToolGui is unsupported, it might have problems now that ExifTool has gone from ver. 8.x to 9.x. Bummer, but I suppose "it was good while it lasted".

BTW: There is newer version of ExifTool.exe, as you may know. Does that make any difference (probably not would be my guess)? Phil Harvey does crank out updates very often ... and I guess that is a good thing.

After unzipping the 5d2 NightlyBuild and the [downloads\20130913_5d2_Modules_f5c083059e9b].zip, I was somewhat surprised to see the file sizes of each of the .mo modules were different. My expectation was that they were the same files with the same file size. Also, I thought WinMerge would report them to be identical.

Is this expected? Is the difference more than just a time-stamp?

Also, I'm curious about:

• Why one .zip has ALL-CAPS for the file-names, and the other has all lower-case. With Windows o/s, I suppose it doesn't matter.
• What the .EN files are for in the cameras flash-card after ML has run. They seem to have zero length.
• autoexec.bin files were the same size, but WinMerge reported binary differences. Expected? Timestamp?
• Now that the NightlyBuild.zip has a complete set of modules.mo (including DUAL_ISO.mo), what is the purpose of the .zip from downloads?

.

What has exif to do with dual ISO?

Probably nothing. I'm having trouble getting the Dual_Iso to work, and thought that removing a non-standard manual lens from the equation might simplify things.

Put some dummy values if your software requires them.

Eventually, the images are fed into the PTGui panorama software package. It needs various EXIF values, especially focal length.

FWIW:
http://u88.n24.queensu.ca/exiftool/forum/index.php/topic,5267.msg25496.html#msg25496

Ah, I hadn't been aware of that. My impression had been that using DUAL_ISO always meant that the resolution got cut in half.

Thanks for clearing up my ignorance ... and your patient assistance with a clueless semi-newbie.

FWIW/TMI: I think I'll retreat to my EF 17-40 uwa zoom for a while so the EXIF is complete. I consider the un-chipped, manual Samyang 14mm to have significantly better IQ, but the missing EXIF may be complicating things.

The shadows aren't that dark.  100/400 would have been fine for this shot imo.

I'm just starting to get DUAL_ISO to work, now that the nightly build for the 5d2 includes DUAL_ISO.mo (thanks!) There does seem to be "some light at the end of the tunnel".

However, my initial impression was that DUAL_ISO provided 4 EV improved dynamic range, using 100/1600, which seemed worth the significant loss of resolution.

I'm still playing around with ACR 6.7 (PV2010) and LR 4.4 (PV2012) to try to get improved images from the .dng's generated by cr2hdr ... slow progress. Seems like PV2012 may be much preferred.

So far, I'm unclear if losing half the resolution is worth the tradeoff to gain "only" 2 EV using 100/400.  For me, "the jury is still out". YMMV.

FWIW/TMI: another issue is I'm using Samyang manual, unchipped lenses (14mm and 8mm fisheye for interior 360 degree panoramas) which leave out EXIF for f-stop and focal-length. Getting some floating-point-overflows when using ExifToolGui (now unsupported), which isn't helping.

[Magic Lantern Best Practices:]

Magic Lantern Best Practices:
http://www.magiclantern.fm/bestpractices

Does a link to "Best Practices" actually exist, or has "bit rot" happened? I'd be interested in looking at a "best practices" thread / article if it exists (not just for T3)

I did a search for "best practices" and "bestpractices" and only got a hit on a T3-related thread above.

Something to consider for cr2hdr.exe:

• enhance so it is a better behaved command line executable
• accept /? to provide help / hints (or even mini-manual to explain usage)
• accept /v for verbose output, as applicable
• accept /q for quiet output
• accept *.cr2 wildcard (now accepts one file at a time from the command line, also "for loop")
• not stop if *.* includes .jpg's
• allow > of output to create log (now works for one file from CmdPrompt window)
• date-stamp and/or version number

Edit:

• This "for loop" seems to work fine from CmdPrompt window:  > for %c in (*.cr2) do cr2hdr %c
• If encounters non-dual-iso image (pebkac) that has full EXIF from EF lens, stops with message: "Doesn't look like interlaced ISO" and the "for loop" continues
• If encounters non-dual-iso image (pebkac) that doesn't have full EXIF from manual Samyang lens, stops with message: "Interfacing method not supported. ISO blending didn't work" and the "for loop" continues

BTW: would be nice if cr2hdr somehow could take advantage of multiple cores, instead of being single-threaded. I've got a 2600k with 4-cores + hyperthreading, and TaskMgr shows about 12% cpu utilization.

original 2.3 ML  + Sep 4 nightly build  +  6bb97bcbc0f4.zip seems to day's 5D2 latest ML

I tried this approach with two-step firmware, with the latest Sep 12 5d2 nightly build, and then the latest Sep 10 modules ... and may have come close to bricking the camera. Drat.

Camera was locked up, with neither control wheel working. The On-Mid-Off switch stopped working, so I had to remove the battery to get it to turn off. Whew. Got messages about invalid shutdown.

Retreated to Sep 12 nightly build without randomly named [modules].zip file from different date. I'm reluctant to try out dual_iso.mo until the 5d2 nightly-build includes dual_iso.mo  (which may be soon?)

YMMV. This is too bleeding edge for me.

EDIT: For Sep 4 a new file exists: 6bb97bcbc0f4.zip at https://bitbucket.org/a_d_/magic-lantern/downloads
It contains dual_iso module. So:

original 2.3 ML  + Sep 4 nightly build  +  6bb97bcbc0f4.zip seems to day's 5D2 latest ML

Thanks for clarifying some confusion on my part!

I guess the SOP is to look in the location you provided for a randomly named .zip file (usually with 12 letters and digits), unzip it to check for sure it has the file-of-interest, and then put on CF/SD flash card after the nightly build.

Or not?

As of Sep 10, the Sep 4 version of DUAL_ISO.mo appears to be the latest greatest.

BTW, both of the following seem to work:
http://bitbucket.org/a_d_/magic-lantern/downloads
https://bitbucket.org/a_d_/magic-lantern/downloads

I'm still unclear why the 5d2 nightly builds seems to include almost all applicable modules like ETTR.M, RAW_REC.MO, etc, but not DUAL_ISO.MO.

[including]

Just download the newest version of Magic Lantern for the 5D Mark II here.

Thanks!

The Aug 30 file (af839274b1fa.zip) has six modules, including DUAL_ISO.MO
arkaniod
autoexpo
dual_iso
ettr
file_man
raw_rec
(not pic_view)

Seems odd, but the Sep 1 nightly build I attempted to get to work for the 5d2 also has six modules, but not including DUAL_ISO.MO
arkaniod
autoexpo
ettr
file_man
pic_view
raw_rec
(not dual_iso)

Also, I just checked the Sep 4 nightly build, and it also appears to NOT include DUAL_ISO.MO    Or am I doing something wrong?

But I suppose "such is life on the bleeding edge".

Make sure the DUAL_ISO.MO module is loaded in the Modules menu in Magic Lantern, then enable it under the Expo menu.

Sorry for being so slow.

I haven't been able to find a pre-compiled-built DUAL_ISO.MO module for the 5d2. Does this exist? I did look at what I think are "the usual suspects" as far as where such a file could be downloaded. Is it inside some .zip?

Would this rusty desktop C++ developer have to build from DUAL_ISO.c ?

Or is my assumption that a 7d/5d3 module would not work for 5d2 ... incorrect?  Flawed? Should I attempt to track down a DUAL_ISO.mo and try  it, and hope my 5d2 didn't brick?

I got no search result "hits" from:
dual_iso.mo 5d2 magic lantern download module

You need to download the cr2hdr program for post-processing the .CR2 pictures to DNGs.

OK. Already done. Didn't contain DUAL_ISO.mo (which is probably "as expected" as this is for the desktop post-processing, I believe)

Please read the white paper for Dual ISO or read the FAQ.

I've read the rather long "white paper" from July 14, which I recall starts off being pretty emphatic that DUAL_ISO only works for 7D and 5d3. Also, I don't recall there being much if any content about how to actually get it to work, as far as something like:
* Step 1: acquire DUAL_ISO.MO for model [fill in the blank  ... 5d2 and T3i in my case]
* 2: Get DUAL_ISO.MO on my flash-card and loaded at boot time
* 3: Set option or options in tab (Expo?) to
* 4. Take pictures
* 5. For Windows, drag-n-drop .cr2's into directory with expanded contents of cr2hdr.zip (mis-named cr2hdr.exe)
* Steps 6 thru n ...
* Very likely several steps left out of above out of ignorance / inexperience on my part

I realize that the whole DUAL_ISO capability is very much "bleeding edge" and evolving very quickly.

I also have attempted and re-attempted to read the "Massive increase ... " thread, but it is comprised of 900+ posts, with references to threads of 1500+ posts. (sorry for the whining).

I am especially interested in getting it to work because a definite interest of mine is doing panoramas. With panos, there is typically plenty of resolution, but dynamic range is a real problem. I'm trying to use less and less HDR with panos, and DUAL_ISO looks like it would be VERY valuable to me.

[Postprocessing]

Postprocessing
- For CR2 files: cr2hdr.c (Windows:cr2hdr.exe)   [ [.zip??? ] ]. It requires dcraw and (optional) exiftool in your executable path (they are in the zip). Drag and drop the CR2 files over cr2hdr.exe.

What is the battery life with and without the external monitor ... and with and without ML?  What about 2.3 Stable vs Latest-Greatest-Bleeding-Edge nightly builds?

I'd be inclined to start with the 2.3 Stable release for the 5d2, which has reasonable documentation. My understanding is that ML was started to realize the potential of 5d2 video and audio.

Note that there has been a LOT of new features to Nightly-Builds, but it can take quite a bit of digging around to figure out how the features work.

[Resolved]

I've got the Sep 1 nightly build, and attempted to read thru 37 pages of thread related to dual-iso, with links to a 52 page thread for raw-rec. Mea culpa, I guess I'm slow, but I'm baffled on how to get it to work.

I've got the raw_rec module loaded, and found what seems to be the ML menu item related to dual-iso:
* Video / HDR video (changed from default 100/1250 to 100/1600)

And then?

EDIT:
Almost resolved? See reply #5 of this thread below.  Turns out nightly build for Sep 1 (and others) didn't include DUAL_ISO.MO

As of Sep 10, it appears none of the nightly builds include DUAL_ISO.mo, for any of the DSLR's I checked (5d2, T3i, 5d3). However, for the 5d2, there will be a zip file with filename consisting of 12 random letters and digits that so far has contained all or most modules, including DUAL_ISO.mo (thanks tron). Look at the following link:
http://bitbucket.org/a_d_/magic-lantern/downloads.

Resolved: 5d2 NightlyBuilds now include latest/greatest DUAL_ISO.MO  (thanks!) 
.zip from /downloads not necessary (?)

Wow, not a single sample from 5D2?!

HA!.... i was right! Dual Iso running on my 5Dmkii... whooo,

is there a compiled 5d2 dual iso module I can test?

Can you explain how to use Dual ISO ...

Great job on Auto-ETTR, and the subsequent revisions to incorporate "Ignore highlight %" and "Clip None/Green/Any".

And again in the spirit of "let no good deed go unpunished" ...

Would it be possible to have an option that tweaked one or several ETTR related settings to be a "best guess optimal" to work well with ACR/LR's PV2012?

My limited understanding is that for least noise, and near-optimal dynamic range when working with PV2012, you want to find that exposure that blows out a certain number of pixels of one channel. Then, PV2012 can make a "best guess" for that channel, based on the other channels. Also, in many images, it may be ok to have 0.1% to 5%+ of the pixels be blown highlights, if the photographer is aware that the scene contains "specular highlights" like the sun, chrome bumpers, light bulbs, etc.

Or perhaps such capability would entail too much ... "it depends"?

Alright, now you have true RAW histogram and zebras. 5D2 only.

Very cool. Great work.

I remain Very Impressed that ML was able to "exceed expectations" on RawDigger-like info, leading to Raw Histograms, Ettr-Hint?, Auto-ETTR?, RawVideo?, whirled peas, etc.

And in the spirit of "let no good deed go unpunished" ...
Would it be feasible / appropriate to implement A)utomated H)DR + E)TTR together? The "user interface" would be how many activations in the bracket, and AHE would take the existing settings for ISO, f-stop, clip choice (green, none, any), ignore highlight choice (0% to 50%), etc., and determine the appropriate shutter speeds.

Or the end-user would supply the f-stop gap, and the AHE would determine the starting shutter speed, and number of brackets?

Note:
To me, ETTR is less of an issue with HDR, as you expect to have one or more images that are underexposed (blocked shadows but great highlights), and one or more that are overexposed (clipped highlights but great shadows). However, it can take time and/or trial-and-error to figure this out.

I would appreciate consideration of having a bracket / HDR sequence that was the same as an alternate sequence that I prefer on the Canon menu:
- (underexpose)
0 (medium exposure)
+ (overexposure)

With that sequence, the middle / medium exposure is likely to be about right, and the "triplet" will consistently go "dark to light". With panoramas, especially 360 degree FOV, I find that simplifies keeping track of separate "triplets".

And ... "while you are at it" (famous last words), how about giving some thought to "auto - ettr - hdr"? (separate feature request)

The autoexec.bin in the nightlies is for all cameras.  The one from the raw recording builds thread is specific to your camera.

Ah, didn't realize that.

That sounds sensible to me.  Why should it ignore the settings you have chosen for your tolerance to overexposure?

The issue was that if the auto-ettr module wasn't installed, how would the histogram ettr-hint get the related/coupled values (for % ignore and clip-what? My speculation is that it would use defaults.

Why has it been moved to Modules?

I'm wondering the same thing. It seems like this will make Auto-ETTR harder to use, so it will get less testing.

Also, my observation is that ETTR-Hint is coupled with the selections for Auto-ETTR, as the selections for "% ignore" and what to clip influence how many EV stops are recommended for exposure adjustment.

OK. Thanks for the hints, and your patience with this ML semi-newbie.

I was able to find ... a43e030494e4.zip,
http://www.magiclantern.fm/forum/index.php?topic=6362.msg50149#msg50149
which contained
AUTOEXEC.COM
ML
ML\MODULES
   5D2_212.SYM
   ETTR.MO
   FILE_MAN.MO
   RAW_REC.MO

And it seems to work. I suppose I can remove FILE_MAN.MO and RAW_REC.MO if those aren't of interest.

I am unclear about the duplicate AUTOEXEC.BIN. The autoexec.bin from the June 19,2013 NB was about 3.1MB, and the AUTOEXEC.BIN from a43e030494e4.zip was much smaller, about 430KB.

Is this the correct "best practice" ...
* Copy the latest NB to the 5d2 CF card
* Firmware update
* Copy contents of unzipped a43e030494e4.zip to CF card, including replacement AUTOEXEC.BAT
* Don't do another firmware update???
* Activate module(s)

I suppose I should check if modules get updated?

Auto ETTR in now a module (like raw rec). 

Sorry to be slow, but I don't find anything that looks like "raw rec" or "auto ettr" in the June 18th .zip file. I looked at other daily entries for "archive" after June 12, 2013, but they were empty. I looked in the 2.3 documentation, but didn't find anything about "module installation".

The module has to be copied to the module folder on the ML installation card

I don't see a "module folder" on the ML installation card. Is that something I create as a subdirectory under ML? Then put "something" in it?

I did create a "module" subdirectory under ML, but that didn't seem to accomplish anything.

and loaded through the module menu in ML.

> Go to modules (rightmost tab in the ML menu) and click load... then the raw video option will show up in the video tab.

The above from another post was all I could find about installing modules. However, my 5d2 and T3i ML menus have "Help" as the rightmost tab.

5d2 and 600d

I replaced 2013-June-11 nightly build with 2013-June-18 NB, and can no long find Auto-ETTR functionality (or other ETTR capabilities like "Hint").

Did I do something wrong? Something awry? Relocated?

Also, I noticed that the June 18 .zip download was quite a bit larger than previous .zip's. Also, that there weren't nightly builds for several days after June 12th.