600D: ETTR and Expo limits reached

[escho]

Hi

I enabled ETTR for my 600D for playing a bit with. It seems to work quite well. But there is one thing, I don´t understand:

If I shoot a scene under normal light conditions( ISO 100, 1/60s), which ETTR can calculate, all is fine. Then I´m shooting a very dark sample and ETTR sets ISO and time to max (ISO 3200m 30s). And it tells me: Expo limits reached.

No problem till now.

But all pictures; i try to take from now on (back under normal light conditions) are heavily overexposed and I always get the message: Expo limits reached. I have to set back ISO and exposure-time manually away from the max to normal values for making ETTR work again correctly.

This happens in both, LV and Photo-mode. ML compiled incl. commit b2bcf64

What´s wrong?

Edgar

[1%]

It seems to be doing it on other cams too... I point it at a bright light on EOSM and it adjusts from too bright to too dark and settles on overexposed as hell or just does dark/bright over and over.

On 7D too it doesn't "breathe" so much but can make too over or under shots quite easily.

[ToniX]

I've read another topic where ity is said that ETTR is not working at all on 600D, hence removed from the latest ML version for 600D: is it back?

[escho]

Hi

I can use ETTR, because I change the code and compile it new. But in practice, I don´t use it, because it´s too slow for the things, I´m shooting. Some weeks ago, I created a pull request to ML to enable ETTR on 600D again. I don´t know, whether this pull request will be accepted or not.

Edgar

[a1ex]

In photo mode it shouldn't have any issues in low light (I've used it in some dark caves and it converged in one iteration - from a 1-second test exposure to a 20-30 seconds ISO 100...800 for final shots).

In LiveView it has this problem. It's very hard to meter a 30-second picture from a 1/30s one with line skipping (though I have some noise analysis in mind that should help here). FPS override might help too.

Can you upload a video showing the problem?

[1%]

Its working on my end and the fix is in pull requests, some commit deleted the state object where raw data was captured.

[escho]

...
Can you upload a video showing the problem?

I will do so, but it will take some weeks

Edgar

[a1ex]

Okay, maybe somebody else can reproduce and upload it faster.

[escho]

Sorry, but I cannot reproduce this fault at the moment.

Edgar

[ShootMeAlready]

Sorry for posting on old thread, but cant find any info. on AETTR for the 600D.  How can one play with it?  I dont see it in the nightly build jan/14/15

[dmilligan]

check your modules menu