How has Dual-ISO changed your shooting style?

[l_d_allan]

Been messing around with Dual_iso raws for the past week, this totally changes my shooting style! THANK YOU!!!

I've been using Dual-ISO for some months, and have been impressed also. And I'm also THANKFUL!!! to the devs.

I'd appreciate getting some details on changes you (and others) have made to your shooting style due to Dual-ISO. You may have "use cases" I'm unaware of, or my use of Dual-ISO may be less than "best practice".

• Mostly I use Dual-ISO as an alternative to bracketed HDR
• Especially for moving subjects, where bracketing isn't appropriate.
• Or when I don't have a tripod.
• BTW: this thread has very useful info to combine ML's Dual-ISO and Auto-ETTR:
   AETTR + DUAL ISO: The Ultimate Automated Perfect Image Exposure-Beginners'Guide

[Audionut]

I'm in the process of creating quite a significant topic about ML features, including dual_iso

I was using dual_iso the other day in overcast conditions.  It was a birthday party under a roof with open sides.  I set my base ISO for the sky (I hate blown highlights), and set the recovery ISO just high enough to not blow wanted detail under the roof.

In this situation, I retained detail in the highlights, and reduced noise in the skin tones/shadows with the recovery ISO.  Didn't have to touch exposure settings all afternoon except for when I wanted to change DOF.  Heh 

Exposure +3.33 - Shadows +62 - ISOs 100/800 - Color noise reduction only.


Previously, I would have been constantly changing exposure based on the background detail and dealing with increased noise in the skin tones/shadows due to my preference of keeping highlight detial.

[l_d_allan]

I'm in the process of creating quite a significant topic about ML features, including dual_iso

Very, very much looking forward to it. Thanks.

Here's a tutorial-like approach that I think would be very informative, although quite a bit of work:

• Attach an image at the "Base ISO" (usually 100?) with Dual-ISO disabled
• Attach an image at the "Recovery ISO" with Dual-ISO disabled (usually 800 to 1600?)
• Maybe? an image shot at "Green / Full Auto" camera settings that may convey the challenges presented by the scene
• List the ML settings related to Dual-ISO you used, and why
• Why you decided this scene was a "good candidate" for Dual-ISO (and/or A-ETTR?)
• Attach an image with Dual-ISO enabled, with minimal or no post-processing
• Describe the Adobe ACR / Lightroom settings you used, and why (or software other than Adobe?)
• Attach an image of the post processed image. This makes it "start to finish" for the entire "tool chain"
• Some hints as to what to look for in the final image to see the difference(s), and what may be too subtle to be obvious.

There are a number of examples in the 1700+ post development thread on Dual-ISO. However, imo that massive thread isn't oriented to being a tutorial.

• http://www.magiclantern.fm/forum/index.php?topic=7139.0
• Dual-ISO has evolved quite a bit since the OP on July, 2013, so usage has changed more or less significantly. Early examples may be obsolete?
• Here's an interesting before/after example: http://www.magiclantern.fm/forum/index.php?topic=10111.msg96377#msg96377

[a1ex]

List the ML settings related to Dual-ISO you used, and why

+1, I could use this data to fine-tune the ETTR + Dual ISO algorithm. The more examples with detailed analysis, the better.

Dual-ISO has evolved quite a bit since the OP on July, 2013, so usage has changed more or less significantly. Early examples may be obsolete?

Will try to update the early examples (also showing them side-by-side with the first version) when I get home.

OT: the idea of including the Dual-ISO tag in EXIF was a nice one; now you can search for Dual ISO shots on Flickr: http://www.flickr.com/search/?q=%22Dual-ISO%22&m=tags

[Audionut]

Hmmm, not sure if it will be that detailed about dual_iso.  I expect the topic to be around 5 or 6 posts just with the detail I already have in mind.

With dual_iso, I am more focused on explaining how it works, so I'll be touching on subjects such as exposure, noise, ISO, ETTR.  Explaining the difference between scientific DR vs photographic DR with practicable examples of the pros vs cons.  ie:  Scientific DR says you gained 14+ EV of DR with 100/1600, but what actual pros did you get in the exposure, and what were the cons.

I'll try and explain the difference between using a recovery ISO for the midtones vs the shadows.

List the ML settings related to Dual-ISO you used, and why

I'm not sure I fully understand, can you elaborate?  For me, dual_iso is simply exposure * 2.

• Why you decided this scene was a "good candidate" for Dual-ISO (and/or A-ETTR?)

Well if you are happy to use AETTR, then this will do everything for you    Being able to decide when a scene is a good candidate requires an eye for exposure imo.  It's being able to look at a scene an discern the DR of the scene.  Ok, you might not be able to say "this scene has 11.5 stops of DR", but you should be able to tell when you need to use dual_iso if you want clean shadow detail.

Setting the recovery ISO is more complicated.  I'll attempt to explain a number of ways to set the recovery ISO with examples.

[l_d_allan]

OT: the idea of including the Dual-ISO tag in EXIF was a nice one; now you can search for Dual ISO shots on Flickr: http://www.flickr.com/search/?q=%22Dual-ISO%22&m=tags

I've been able to guestimate the "base" and "recovery" ISO from the value of :
ISO difference: #.##
that shows up in the output of cr2hdr.exe (or the .log files from the .vbs multiprocess script)

100/800 ~= 3 +/- 0.2
100/1600 ~= 4 +/- 0.2
100/3200 ~= 5 +/- 0.2

Or not?

Feature request: it'd be nice to have cr2hdr.exe output / log / printf / debug something that would facilitate identifying base/recovery ISO's

[l_d_allan]

+1, I could use this data to fine-tune the ETTR + Dual ISO algorithm.

Different people have different "use cases". My speculation is that ML users are several sigmas outside the norm.

The more examples with detailed analysis, the better.

For example, I really struggled with photos of a play with spotlights on a darkened stage back in December,
http://berean.zenfolio.com/dual_iso/h5db3b7b#h5db3b7b
http://berean.zenfolio.com/dual_iso/h5db3b7b#hcaf32f3

so I attempted to rig up a "synthetic equivalent" to experiment with:
http://berean.zenfolio.com/dual_iso/h5db3b7b#h1890b735
http://berean.zenfolio.com/dual_iso/h5db3b7b#h161de120

It's got a Datacolor SpyderCube in a flashlight's spot, and a D.I.Y. spydercube out of the spotlight, and in the shade from the tripod legs:

And another high-contrast synthetic test, with a stainless spoon for spectral highlights, styrofoam cups for whites/highlights, plus WhiBal card for light gray, black, and white.
http://berean.zenfolio.com/dual_iso/h101028e1#h101028e1

Full moon, urban with significant light pollution:
http://berean.zenfolio.com/dual_iso/h27797ded#h27797ded

My thinking is that a "tutorial bundle" might include:

• a link to a .zip of the original .cr2's and .dng's (which could be a large .zip)
• the .dng's could have ACR/LR snapshots  at different stages (as-shot; auto-exp; samplers used to do by-the-numbers DR on Blacks, Shadows, Mid-Gray, Highlights, Whites; pre-cr2hdr?; post-cr2hdr?; etc.)
• ...
• In addition to ISO 100 and ISO 1600 without Dual-ISO mentioned in previous post, perhaps also have ISO 100 at base shutter speed, and a shutter speed 16x longer. This would sort of simulate bracketed HDR to compare to what Dual-ISO can accomplish.
• Similar for 100/800
• IIRC, the NeatImage plug-in for NR (noise reduction) has the capability of objectively measuring noise in an image. I've barely used the NeatImage plug-in since PV2010 in CS5/ACR6 provided much improved NR.

Will try to update the early examples (also showing them side-by-side with the first version) when I get home.

+1
That might clear up some confusion (and frustration) by non-devs like me attempting to wade through 1700+ posts.

[a1ex]

http://berean.zenfolio.com/dual_iso/h5db3b7b#h5db3b7b
http://berean.zenfolio.com/dual_iso/h5db3b7b#hcaf32f3

This situation can be a great candidate for Dual ISO. If you can freeze the motion at ISO 100, it should be the way to go.

But if you are getting it underexposed at ISO 1600, then just stick to ISO 1600. That's a common mistake from what I've seen in the forums: dual ISO is not going to make the high ISOs cleaner; it works best in strong and contrasty scenes where you can ETTR at ISO 100 (and still get noisy shadows).

I also find it very helpful outdoors in full sunlight (with the sun in front of you, but not only).

[Marsu42]

it works best in strong and contrasty scenes where you can ETTR at ISO 100 (and still get noisy shadows).

Looking at the posts around here, this seems to be a common mistake. Since my "min. dr gain to enable dual_iso" didn't make it into trunk, maybe some other strong reminder about this would be in order, like for example an .advanced option that needs to be toggled to enable non-100 iso values, otherwise (like I did in my local ml build by now) simply force iso to 100 once I dual_iso shooting is enabled... just an idea, works for me.

[a1ex]

It didn't make right now, but I still keep the idea in mind and will hopefully find a simpler solution that doesn't require extra care from ETTR.

Probably a piece of advice in the help text (just as it tells you to avoid 100/6400) would be good. Especially if somebody tries to do 800/3200, for example.

[engardeknave]

I was thinking about attempting dual ISO HDR. I haven't bothered to figure out how many stops I'm gaining with 100/1600. But that would certainly reduce shutter actuations.

[l_d_allan]

I was thinking about attempting dual ISO HDR. I haven't bothered to figure out how many stops I'm gaining with 100/1600. But that would certainly reduce shutter actuations.

I'll be interested in your evaluation.

For me, I think of Dual-ISO as a way to avoid many of the HDR's I would have made in high contrast situations. I'm trying to use HDR less and less, especially with panos, where there is plenty of resolution. I've been very pleased and impressed with Dual-ISO, although I doubt I've gained experience to be confident I'm using "best practices".

As far as combining HDR + Dual-ISO, my speculation is that might not be all that productive. As long as you have decided to do HDR bracketing, I would think the use of HDR might save you one shutter activation. My norm for HDR tends to be 3 images at 2 stop bracketing.

Perhaps, the use of combining HDR + Dual-ISO might reduce this to 2 images. You've still got the extra workflow for HDR. YMMV.

[l_d_allan]

Feature request: it'd be nice to have cr2hdr.exe output / log / printf / debug something that would facilitate identifying base/recovery ISO's

Disregard ... I looked closer, and noticed that the effective actual ISO was in (parenthesis) next to the ISO difference value, such as:
ISO difference  : 2.05 EV (413)
ISO difference  : 3.04 EV (823)
ISO difference  : 4.03 EV (1631)