Auto-ETTR feature requests

[Marsu42]

I've been trying auto-ettr a lot today, it's really a terrific feature, great work! The one real limit is the need to take one trash pic or enable lv, but of course than cannot be helped. I have some feedback since I feel this great feature deserves to be as usable as possible for as many people as possible:

1. the unavoidable request: Please add an option to allow ettr to open the aperture to user settable limit (so the dof doesn't get too small). Ideally there would be a priority option what to do in what sequence: raise iso (until Canon limit), lower shutter (until set limit), open aperture (fixed like now or until set limit). Rationale: If looking for a proper exposure for example on my 17-40L I'd rather go from f8 to f5.6 (wide open would be f4) then raise iso further, and if auto-ettr doesn't do it for me I have to do it manually.

2. DONE auto-ettr doesn't seem to work with "image review" off? If so, ml should set it to 2s to avoid confusion EDIT or at least give a warning message in the menu.

3 EDIT DELETED because you can overexpose the 1st shot, but have a 2nd to get the 3rd to ettr.

4. DONE option to prevent 2nd shot before ettr calculation: I stood at a very busy street in bright sunshine and could neither hear the ettr ready beep nor see the display. Sure I could just wait some, but it would be nice to have an option to block the 2nd shot until the ettr calculation is done, otherwise the 2nd is like the 1st... if blocking is not possible maybe an option to do the 2nd shot automatically once ettr calculation is complete?

5. EDIT DELETED because ml cannot re-load previous shots as raw and ettr them.

[Audionut]

1.  http://www.magiclantern.fm/forum/index.php?topic=5200.msg43395#msg43395
2.  This could be handy I guess.  Is image review on? No.  Enable 2 sec.  But when to set it back to original setting?
3.  Learn exposure.  Any of the modes other then M should be giving you a technically correct exposure, and a good exposure for ETTR.
4.  People shouldn't be limited in their ability to take an exposure just because others need to rely on features.
5.  Av, Tv, P modes should be getting you an exposure that's technically correct, and good for a following ETTR exposure.  In M mode, don't change the settings!  Not possible to reload Raw data.

[Marsu42]

2.  This could be handy I guess.  Is image review on? No.  Enable 2 sec.  But when to set it back to original setting?

2. Restoring settings ml changed is often tricky (like when the camera is turned off...), so the user will have to live with this - the alternative is to print an error message to let the user change the setting him/herself, but only in lv & not in photomode w/o img review, so in this case I'd just change img review to 2s.

3.  Learn exposure.  Any of the modes other then M should be giving you a technically correct exposure, and a good exposure for ETTR.

3. What have the other modes got to do with it - I need to do the 1st shot in M to get a ettr exposure? If it would be possible to take the 1st shot in Av/Tv and then switch to M this would still involve turning the dial, but also a nice addition. PS: Please don't patronize me, thanks.

4.  People shouldn't be limited in their ability to take an exposure just because others need to rely on features.

4. Hugh? I don't understand what you're saying - Do you mean improving a feature is not necessary because people would rely on it if it worked all the time?  I feel that the ettr ready feedback is too subtle in certain conditions or when you don't want a beep, so blocking/automatic the 2nd shot would be a usability enhancement.

5.  Av, Tv, P modes should be getting you an exposure that's technically correct, and good for a following ETTR exposure.  In M mode, don't change the settings!  Not possible to reload Raw data.

5. Thanks, I already suspected reloading raw data is impossible

[Audionut]

2. so the user will have to live with this - the alternative is to print an error message to let the user change the setting him/herself, but only in lv & not in photomode w/o img review, so in this case I'd just change img review to 2s.

The user could just change the setting himself so as not to implement code that could upset other users.
There's heaps of other ML options that require user intervention.  IMHO, this is another one of the many options that requires the user to ensure they have the correct settings before proceeding.

3. What have the other modes got to do with it - I need to do the 1st shot in M to get a ettr exposure? If it would be possible to take the 1st shot in Av/Tv and then switch to M this would still involve turning the dial, but also a nice addition. PS: Please don't patronize me, thanks.

My mistake, all this recent talk with you regarding all these other modes I don't normally use 
There's nothing patronizing about learning exposure.  It would solve this problem.

edit:  You don't have to underexpose by 13 stops for ETTR to work.  It will work even if the initial exposure is overexposed (although it will take longer to find the optimal result). 

ETTR actually taught me how to take an exposure in manual mode quickly  (it was very daunting before this).  I like the technical stuff, so I quickly found where on my cameras exposure meter (spot meter mode), over exposure occurred.  Quickly look for the brightest part of the scene and put it at +2 and 2/3rd's EV on the exposure meter. 

Skin?  Put it at +2EV on the exposure meter (leaves some headroom).

So you don't have to waste time, waste an exposure, kill the shutter with AutoETTR.  With a little thought and practice, you can get your first exposure better then the camera can anyway.  AutoETTR (to me), simply means that all subsequent exposures will be the perfect ETTR shot.

4. Hugh? I don't understand what you're saying - Do you mean improving a feature is not necessary because people would rely on it if it worked all the time?

No.  I mean, why should there be a delay in exposure capture with ETTR on?  You're telling me it's because in a certain situation you fired off some shots to quick.  It's got nothing to do with being able to see the screen or hear a beep.  The calculation takes a sec.  Give it a sec.  Someone who has ETTR enabled but wants to take rapid exposures regardless, shouldn't have that ability taken away from them with some delay in code, just because you feel that ML code should revolve around how you shoot.

I'm honestly not being patronizing.  I'm taking the needs of other users into account!

edit:  I guess it could be an option (delay), but IMO, options should be limited to a certain extent so as to not clutter the GUI.

[Marsu42]

The user could just change the setting himself so as not to implement code that could upset other users.
There's heaps of other ML options that require user intervention.  IMHO, this is another one of the many options that requires the user to ensure they have the correct settings before proceeding.

On reflection, you might be correct - but there should at least be a (yellow) warning line in the menu when enabling ettr with image review off because the connection really isn't obvious.

edit:  You don't have to underexpose by 13 stops for ETTR to work.  It will work even if the initial exposure is overexposed (although it will take longer to find the optimal result). 

Indeed, from the description I wouldn't have expected that thus I didn't try it, thanks! For severe overexposure it needs 2 test shots though if the whites are clipped in raw in the 1st one, I'm much better knowing that though it kills the shutter.

Quickly look for the brightest part of the scene and put it at +2 and 2/3rd's EV on the exposure meter.  Skin?  Put it at +2EV on the exposure meter (leaves some headroom).

It's amazing how much dynamic range you can get out of raw, though of course your values will be at iso100 where the dr is largest? Before ml ettr I though maybe 1-1.5ev overexposure would be max as I have done for years, I really never really tested it scientifically. But of course manually ettr'ing always risks blown whites and esp. specular highlights if you were wrong :-o

No.  I mean, why should there be a delay in exposure capture with ETTR on?  You're telling me it's because in a certain situation you fired off some shots to quick.  It's got nothing to do with being able to see the screen or hear a beep.  The calculation takes a sec.  Give it a sec.  Someone who has ETTR enabled but wants to take rapid exposures regardless, shouldn't have that ability taken away from them with some delay in code. edit:  I guess it could be an option (delay), but IMO, options should be limited to a certain extent so as to not clutter the GUI.

When getting used to how much time ettr takes I'll be more acustomized to it even w/o the need of camera feedback. But I still feel that a delay option would be a good addition, though maybe not a essential one - it's up for the devs to decide that after collecting feedback.

just because you feel that ML code should revolve around how you shoot.  I'm taking the needs of other users into account!

You're getting this completely wrong, and I should explain it: My feedback here is *deliberately* based on *my* impression because chances are that what I discover other users share - and most of the time I posted some request here some other user(s) agreed that that's what they also miss or think but simply didn't convey. Or there will be a discussion and a better/modified way emerges.

If I'm all alone with a feature request it's completely ok to ignore it, if pressed I can program it myself in my local ml version. But "taking the need of others into account" simply only works up to a point because people are so different, that's why imho feedback is so important esp. in the dev phase when things are about to be changed often anyway.

I'm honestly not being patronizing.

No problem, just wanted to mention that "use the green square mode", "learn exposure" and "a little of the grey matter is needed" might be received otherwise as you have intended it :-) but maybe I'm overly sensitive, and of course communication styles are always very different, shouldn't stop us from exchanging ideas, testing and trying to improve ml.

[Audionut]

It's amazing how much dynamic range you can get out of raw, though of course your values will be at iso100 where the dr is largest? Before ml ettr I though maybe 1-1.5ev overexposure would be max as I have done for years, I really never really tested it scientifically. But of course manually ettr'ing always risks blown whites and esp. specular highlights if you were wrong :-o

Dynamic range is limited by the shadows (noise floor).  Take a plain wall, spot meter it @ ISO 100 and increase exposure until it starts to saturate the pixels (overexpose).  Take note of where the Canon exposure meter reads at saturation.  Increase ISO to 1600 and expose the wall at the same EV reading on the Canon meter

Edit for clarification: Highlight headroom (0EV on the meter to maximum pixel saturation) remains the same for all ISO's.  All bets are off for the digital amplified ISO's.   Although the headroom should remain the same, it's at the discretion of the developer when coding the digital gain.

You're getting this completely wrong, and I should explain it: My feedback here is *deliberately* based on *my* impression because chances are that what I discover other users share

No problem, just wanted to mention that "use the green square mode", "learn exposure" and "a little of the grey matter is needed" might be received otherwise as you have intended it :-) but maybe I'm overly sensitive, and of course communication styles are always very different, shouldn't stop us from exchanging ideas, testing and trying to improve ml.

You are correct.  I'll continue further via a pm. 

[a1ex]

2 was already done since the first implementation...

4 done, and seems to be a nice time saver, since it takes the next picture before quick review pops up on the screen.

- If the exposure is spot on (within a small tolerance), it takes one shot.
- If it's slightly underexposed (say less than 5 EV or so), it takes 2 shots.
- If it's overexposed, it usually takes 2-3 shots.
- If it still can't get it right after 4 shots, it gives up.

Bonus: it can be used like a simple bracketing, sort of.

[Audionut]

Wow.  This works good, and fast!

edit:  Only problem being, we've gained speed and ease of use, but lost the precision.

With 0% ignore, ML is settling on exposures with maximum pixel saturation values of 8875 as shown by RawDigger.  RawDigger reports the maximum pixel saturation of my camera at around 13200.   Previously, ML would have made the exposure much more closer to maximum pixel saturation.  Currently it's 2/3rds of a stop below saturation.

With 0.3% ignore, ML is settling on exposures with maximum pixel saturation of 12100.  Close to saturation, but previously, ML would have blown a small percentage of highlights, as should be expected.

[a1ex]

Note that maximum target level for ETTR can only be -0.5 EV or less. If I push it to 0, it will have no way to know how much it should go back if it overexposes a little bit, so the algorithm convergence will suffer.

Also, the accepted tolerance in ML is between -0.2 and 0.7 EV, so it has a preference towards underexposure (you said it's better to underexpose a while ago). I didn't think it's worth wasting another click for such a small difference.

Another source of errors: at ISO 100, ML is only constrained by shutter rounding (only the following fractions are accepted by Canon code: full stops, 3/8, 4/8 and 5/8 EV). At higher ISOs, the ISO can be set in full stops, and shutter has to be ideally the value from menu. 1 EV of rounding error is a bit too much, so I'm using 0.5 here.

[Marsu42]

2 was already done since the first implementation...

It is? Well, on my  60d it isn't working - if I set img review to off ml doesn't ettr and doesn't turn img review on.

4 done, and seems to be a nice time saver, since it takes the next picture before quick review pops up on the screen.

Great, works fine :-) plus the bracketing is a good idea since if it was a high dr scene in the first place. Also the renamed menu item is more self-explanatory now.

Also, the accepted tolerance in ML is between -0.2 and 0.7 EV, so it has a preference towards underexposure (you said it's better to underexpose a while ago). I didn't think it's worth wasting another click for such a small difference.

+1 ... and a little safety margin is good even with ettr if it's not a completely static scene.

[a1ex]

Look in menu, there are warnings... didn't try on 60D though.

[Marsu42]

Look in menu, there are warnings... didn't try on 60D though.

Ok, sorry, I though you'd said that ml switching img review on was implemented, the warning is there - so forget about it :-p

[Audionut]

I'm being pedantic.

With 0% ignore, it should never saturate pixels (which it doesn't).  It works great thanks a1ex.