Bulb ramping

[tomms]

Hey there, ive been using magic lantern on my 50D for about a week now with bulb meterting.  Im having issues with the ramping on sunsets

-i take a well exposed shot
-Magic lantern analyzes it
-I meter for the midtones at the 50 percentile

But then i get these huge jumps. For example, it will go from 1 to 3 second exposure.  Also it seems as though only the exposure is ramping, the iso stays the same.

Can anyone shed some light on this?

[Francis]

The first few shots will jump around a lot until the algorithm finds its way back to the intended exposure. This is where the exposure curve smoothing comes into play. Also depending on what your initial shutter speed is, the precision of the bulb times varies depending on the model. Not sure specifically about 50d. The shorter the initial shutter speed is, the greater variation in exposure when the actual length of the bulb exposure is a little off from what ML calculates the exposure should be.

As far as the ISO not changing, the algorithm shifts ISO as it approaches the length of the interval. So say your interval is 8s, when the necessary shutter length is 6s, ML will bump up the ISO by 1 stop to bring the shutter down to 3s. This gives some headroom if the exposure changes drastically.

I'm not sure if these exact values fit the algorithm now, as it has changed substantially since I last worked with the code. Previously, ML aimed to keep the shutter speed between 90degrees and 270degrees of the interval. But whatever the algorithm is now, the idea will still be the same, relying on ISO change only when it starts to get close to the interval. Keeping the ISO low keeps the noise down and image quality up.

Please someone correct me if I'm wrong. As I said the bulb-ramping feature is been greatly improved and is much more complex than 6months ago when I was trying to wrap my head around how it worked by looking at the code.

[a1ex]

Large jumps appear when the ramping speed selected in ML menu is much lower than the real one (too much smoothing).

[snoopstah]

Thanks for this, I've been having exactly the same issues with my 50D, which I've been doing tests with for the last two nights.  I will try a larger ramping speed tomorrow.

I assume there is also a downside to setting the maximum ramping speed too high?  Is there a way to determine approximately what the optimal setting should be?  Presumably on a clear evening it is largely related to your latitude?

[a1ex]

Post the BRAMP*.log file (under ML/LOGS) and I'll show you a graph from where you can read the ramping speed.

If it's too high, the flicker will be a bit higher, but nothing you can't fix in post.

[garry23]

As another 50d owner I also hae been experimenting with bramping in 2.3 and like others have come across the 'setting the right parameters' problem.

I hope someone in the know can provide some advice regarding bramp speeds. I have read elsewhere that at sunset the ev will change about 2ev per 10 min. Maybe this gives a guidance figure for speed bramp?

Bottom line looking forward to some guidance.

Ps have any of te 50d owners worked out how to move the cursor up and down the calibration curve. When I use te small wheel it does not move very far, ie I can not get it to move fully up and down te curve.

Cheers

Garry

[snoopstah]

Post the BRAMP*.log file (under ML/LOGS) and I'll show you a graph from where you can read the ramping speed.

If it's too high, the flicker will be a bit higher, but nothing you can't fix in post.

Thanks so much for your help.  I tried a significantly higher ramping speed last night (0.5EV I think) and got much better results, and also found that, when comparing the BRAMP log file with my previous efforts, I no longer have any log entries for 'harsh' adjustments, where the exposure suddenly changes.

I have posted the BRAMP log for last night's run (my best so far) here:

http://pastebin.com/gprQ4zdh

If you are able to give me some tips on how to use this to determine an optimal ramping speed, that would be very much appreciated.

Also, I have another question, if you have time.  Looking at my photos from last night, there is a significant jump in brightness at file 5925, and looking in the log file you can see that reflected in the 'luma' column as well.  This takes around 40 frames to fully correct.  My understanding is that this is because this was the point when the camera moved from using a fixed shutter time (1/10th) to bulb mode controlled by ML, and that the 50D probably is unable to accurately do bulb mode exposures this short?  Is this a reasonable assumption?  Is there any way to adjust when ML switches from using fixed shutter speeds and moves to bulb mode, short of recompiling?  It would be interesting to test what happens if I use fixed shutter speeds down to, say, a third of a second, before then transitioning to bulb mode.

Many thanks for your help and for your excellent work on ML!

[Milutiche]

Im also having trouble with bulb ramping sunrise for timelapse, is there a good tutorial somewhere, easy to follow for a noob.

thanks, Jason

[a1ex]

@snoopstah: Here's the graph:



Can you post the exif info for last non-bulb pic and first bulb pic? The difference seems to be about 1 stop.

Ramping speed was 5 stops / 120 pics, so 0.04 stops / picture.