Bulb ramping Trouble Shooting

Started by RenatoPhoto, January 17, 2013, 04:33:30 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

RenatoPhoto

I am new to this technique and would like to learn and analyze the results of the BRAMP.LOG file.

I have imported the LOGS into excel using fixed width as separators of the fields.

Could someone help me explanain of the following items:

1.  Luma curve: cached: 11,16,25,43,78,128,184,226,248,254,254
     (I assume that is the set of points used in the internal calibration S curve, in which case it only represents the x-value but no y-value is given in this curve.  I assume y value is stored somewhere.)

2.  Reference level: 69 at 65-th percentile  (I assume this is the percentile I selected from the 1st image I took and to be used for the future analysis)

3. Now there is a set of parameters for each image taken that need clarification:

First Column:   Image Number
Luma = xxx  (Luma calculated from new image, how is this calculated)
rounderr= xx (Dont know)
y= xx (Dont know)
r= xx (Donk know)
e=-xx (Dont know)
==> harsh/soft
f= xx
e= -xx (Same as above)
u=- xx
shutter= xx (Image Shutter speed value)
iso= xx (image ISO value)

Thanks
Renato
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

RenatoPhoto

The second thing I need help is which values from above are used to make a graph as in link

http://a1ex.magiclantern.fm/bramp/rramirez125-BRAMP01.png
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

Here's a description of the algorithm and and here's the script I use for the graphs.

Now, the questions:

1. X is -5:5 EV.

2. See https://en.wikipedia.org/wiki/Percentile . This thingie is statistically robust, unlike plain average => small random objects in the frame will not cause flicker. The 50th percentile is the median.

3. Luma is from 0 to 255, computed at the selected percentile; it's then compared to the reference level. These are converted into EV, and the difference is "e" - the tracking error (see http://en.wikipedia.org/wiki/Control_theory ).

"r" is nonzero only when you use manual ramping (this is the desired brightness, normally equal to the brightness of first picture, which is coded as 0).

"f" is the smooth factor (the location of the poles of the closed-loop).

"u" is the correction applied to shutter speed (it's accumulated).

If shutter becomes too slow or too fast (according to the 180-degree rule), ISO is adjusted too.

Theoretical performance (simulated in octave):










RenatoPhoto

Thanks Alex.  That was really great information.

I am not sure how to open py script.  I created and excel file and printed the information on a graph.  There must be some additional formulas in py to get the graph that you displayed.

So Timelapse flicker heavily at some point.  Some transitions are very good but not all of it.

Cam is 7D, intervalometer at 35sec, AutoExpoRamp = Sunset, Max RampSpeed=0.2 EV/shot

Here is Log http://www.pululahuahostal.com/ML/BRAMP24.LOG

Here is my Chart of information in BRAMP23.log:  http://www.pululahuahostal.com/ML/7D-Auto-35-sec-interval.jpg

Here is the excel sheet with BRAMP24.log:  http://www.pululahuahostal.com/ML/Jan-17th-7D-Auto.xls

Any suggestions on what is needed to improve the TL by reducing flicker.

I hate to use your valuable time but I cannot find the answers.  This is just one of many many tries and I am just not getting rid of the flicker.

Thanks in advance for your help!
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

It was a bug in the algorithm; was fixed after 7D alpha 2. Run the timelapse through my deflicker script and it should be fine.

For flicker-free timelapse, try to use exposure times higher than 1 second. This can't be improved; at faster shutter speeds you will get flicker (but at least you get the raw material to work with).



The problem around 1820 looks very similar to the impulse response from this simulation (compare to the one in previous thread):


If you look close, the difference between the theoretical performance that I was seeking, and the one that I've actually got on the field, was mostly in micro-smoothness, and this is why the bug went unnoticed.

RenatoPhoto

Can you please provide the formulas to calculate the EV graph?  I will put these formulas in Excel for those who not use python. 

I opened the py script but was not able to fully understand the formulas used.

I also ran py under cygwin but it seems that I need to install pylab. I dont want to do this now in case I screw up the installation and later I will not be able to compile autoexec.bin

Thanks for your support
Renato
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

The EV curve is -log2(shutter * iso) + some constant. Or, -cumsum(u).

To get the measured scene brightness, add y to the EV value used for picture at step k-1.

Then reverse the graphs so that higher values mean longer exposures (darker scene).

See also http://dougkerr.net/pumpkin/articles/APEX.pdf .

RenatoPhoto

Thanks, that was very nice.  I enjoyed the reading too!  I am an engineer and appreciate the formulas and a deeper understanding.

I managed to get the EV formula to work and used a constant of 10.5 to get values similar to yours.

Now I am stuck on calculating the measured  brightness.  If I add y to EV I get much larger numbers..
for example on the photo number 1872 y = -103 and EV is -10.6. How to add them?
Thanks
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

RenatoPhoto

Ok, I made the following corrections:

EV=ABS(-LOG(shutter*ISO/100;2)+K .. this formula agrees pretty well (graphically) to the py graph, but I cannot confirm that it is exact since I cant run the py script.  I am using k=4 to get aprox results but dont know if that is the exact number to use.

I assume that this EV is the "Bulb Exposure" from your graph.

Meassured Scene Brightnes = EV + Y/100 ..  Looks pretty similar (graphically) to the py graph., but I cannot confirm that this is the exact equation.  I understand how EV is obtained from the image but I do not undertand where does the Y value come from.  Is this a value calculated by the camera when you half press the shutter??

I assume that "Meassured Scene Brightnes" is the "Estimated scene exposure level" in your graph.

Any help in understanding this is much appreciated.

Thanks
Renato
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

nanomad

You can open the py file with a text editor and extract the correct formula if you wisg
EOS 1100D | EOS 650 (No, I didn't forget the D) | Ye Olde Canon EF Lenses ('87): 50 f/1.8 - 28 f/2.8 - 70-210 f/4 | EF-S 18-55 f/3.5-5.6 | Metz 36 AF-5

RenatoPhoto

As offered, I have created two excel sheets to produce graphs for the BRAMP.LOG

One sheet for 7D (Alpha2)  http://www.pululahuahostal.com/ML/BRAMP_GRAPH-7D.xlsx

One sheet for 5D3 (Unified compile) http://www.pululahuahostal.com/ML/BRAMP_GRAPH-5D3.xlsx

NOTE THAT MY FORMULAS ARE IS SPANISH SO YOU WILL HAVE TO REPLACE THESE FORMULAS FOR THE ENGLISH VERSION.  I HAVE PROVIDED A LIST OF THE TRANSLATIONS IN SHEET 1.  If you make the changes of the formulas then it would be nice if you posted the English version in this thread also.

The reason why they are different is that the 7D log provides a shutter speed in seconds multiplied by 1000 therefore 20 seconds would show up as 20000.  On the 5D3 log it would show as 20 so the calculations are slightly different.

One more trick is required for the 5D3 or unified sheet to work.  In the BRAMP.LOG the shutter speed is in the format xx.xx and my excel program wants a coma instead of a period so you must do a search and replace the entire column of the shutter info.  Search for . (period) and replace with , (comma)  Check your version to see if it accepts commas or periods for the decimals and change the data accordingly if necessary.

1. To open the BRAMP.LOG in excel you need to select all file types (*.*) since excel will not recognize the .log extension. 
2. The import assistance will open and you should click on the "delimited" option, then click on Next, then click on the "space" tick-box, and click on Finish.
3. In most situations the import will be fairly clean but, sometimes I have to fix column C and D because the log does not have a space between the Luma= and its value so it moves all the columns to the left.  Since this is not repeatable you will have to fix this yourself.
4. If using the unified compile you will have to replace the period for a coma in the shutter column.
5. Copy the data from the BRAMP.LOG file to the BRAMP_GRAPH-5D3.xlsx  and pasted it at the first TAB named "BRAMP_Data".  The calculations are done at the "APEX_Calculations" TAB and the graphs are displayed at the next TAB "GRAPHS"

Since this is not an Excel Help blog I will not provide much help so I expect you to know enough about excel to use this sheet.   If you dont know how to use formulas, graphs, etc in Excel then do not even download these files.

If you are good with Python then use the py script provided above by Alex.
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

RenatoPhoto

Upon further analysis of the data and comparing the Log with the Photos I have discovered unusual high errors on the shutter speed.  This always happens just after the camera goes from 1 second exposure to the  fraction exposures.  I say that the errors are unusually high because the shutter speed tends to jump in this area more that I would expect.

All of the analysis is based on the Excel Sheet :  http://www.pululahuahostal.com/ML/BRAMP_GRAPH-5D3.xlsx
And the log: http://www.pululahuahostal.com/ML/BRAMP34.LOG

What I have done is obtained the shutter speed from the photos themselves and compare them with the shutter speeds on the Log file.  I noticed that the log has a photo number x but it appears that this information really applies to the (real) photo number x+1.  Also I am assuming that the Photoshop Raw plug in is interpreting the image shutter speed as close as possible since the logs have a different shutter speed.  Again the closest match between the actual picture and the log is by using the photo =log photo number + 1

Assuming that this is correct I am presenting a graph of shutter error which I interpret as the log shutter and what actually shows on the Raw plugin.

The first image is error versus photo number:  http://www.pululahuahostal.com/ML/Shutter_Speed_Error-Vs-PhotoNumber.jpg

As the photo number goes to where the shutter is below 1 s the errors just shoot above 20%

The second image shows the error vs shutter:  http://www.pululahuahostal.com/ML/Shutter-Error-Vs-Shutter_Speed.jpg

I have also noted that the error continues a bit high as the shutter speed gets shorter but definately there are huge errors between shutter speeds of 1/10 to 1/3 .  These errors show significant flickering which only appear in this section of the Timelapse.

I have read MANY places to keep the Timelapse with bulb mode to reduce flicker.  This is absolutely true and these graphs prove this but...

I have done other timelapses in manual mode where the shutter speed in this range does not cause the amount of flicker that the BulbRamp Timelase with ML produces.  This leads me to conclude that there must be an error in the ML calculations as the system moves from bulb exposure to fractional exposures.

Also by reviewing the actual photos I can see this unusual fluctuations on the shutter speed as we move through this specific point.

The Timelapse produced by the unified ML is excellent until you reach this bulb/fraction point, then flickering begins for a  period and then it seems to settle down again.

I use MSU deflicker and get good results but the errors pointed here cannot be removed by deflickering software.

I dont know enough C code to decipher inner working of shoot.c so I hope that someone with this knowlege can take a look at this area where I think (in my opinion) that the script could be modified to reduce the flicker (shutter) errors.

Renato

http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

I've measured the timing errors for around 0.2s with Audacity and got around 20 milliseconds of variation of the delay between two mirror actuations, in bulb mode. With M mode, the delay variation is less than 1ms.

That's one reason why in recent commits I've switched to M mode for anything under 0.5 seconds.

QuoteI use MSU deflicker and get good results but the errors pointed here cannot be removed by deflickering software.

My deflicker script handles even severe flicker like this:

original picture sequence (2 EV overexposure in the middle):


corrected:


large jpegs: 1 2 3
raw files: 1 2 3

video: http://vimeo.com/48861358

RenatoPhoto

Ok, I have not been able to use the raw deflickering software you provided for 5D3 and the new RawDeflickering package will not install in my Windows machine. 

But..  The main question I have is if the camera can do 1ms repeatability then this would represent a 1% error on a shutter speed of 1/10.  I am measuring error of more that 20% between the raw image and the log file data.  That is what I am pointing out as an error that I don't understand.  I am guessing that there is something fishy in the code that produces these large error when we go below 1 second shutter speed.  A 20% error at 1 ms would be on a 1/200 sec shutter speed but we are getting 20% and even 30% errors between speeds of 1/3 to 1/10.

Remember that what I call error is the difference between the log shutter and the Raw Image shutter.  Also this is something I don't really understand since I though the camera was just taking the photo based on what the software calculates so if the script calculates a 0.2196 seconds shutter speed why is the Raw image come up with 0.33 instead of say 0.2 which would be much closer.  This happens between image 4541 and 4542.  Now considering that 4541 had an error of +26%, then next image with an error of -34% gives a total difference of 50% in apparent brightness change which is very noticeable.  Again this only happens in this range of shutter speeds.

Would it be possible to add a 5 point moving average to the correction as we close to the shutter range below 1s to try to smooth out these errors?  I was watching the camera at this specific point and following the changes manually in my 7D and I could not see why the shutter speed was jumping so much, so what I am saying is that if I was doing this manually I would have just done things slower to prevent the shutter speed jump between images 4540-443.

To illustrate the jumps I put this table:

Image N   Log Shutter    Actual Shutter @ Image N +1
4540          0.2529             0.20
4541          0.2192             0.33
4542          0.1815             0.20

In this particular case the shutter could have remained at 0.2 so why did the camera interpret the 0.2192 as 0.33.  So was there error in the interpretation of the camera or was it the code that interpreted the instruction incorrectly?
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

The windows deflicker will not work for 5D3, only the Linux one does.

I've explained the flicker bug for 7D before (it's solved in recent code). Adding extra smoothing will not help at all (if you set bulb timer to 0.2 seconds and take 5 pictures with the intervalometer, it will flicker).

RenatoPhoto

One more question.

If the log file says that the shutter should be 0.2192 why does the camera shoot the photo at 0.33 shutter speed instead of say 0.2?  This is the thing I call an error and I dont understand.

And thank for all of your replies.  I really appreciate the help.

Renato
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

From the moment when ML issues a shutter command, until Canon code actually processes it, there is some delay (there are lots of tasks processing this request, it goes through some message queues and it ends up in the MPU, probably the real-time CPU). There is no exact control over this timing; I have only measured the repeatability of the process. Each camera may require a different bias though.

I have no idea how accurate is the exif info.

RenatoPhoto

Quote from: a1ex on January 25, 2013, 12:48:36 AM
From the moment when ML issues a shutter command, until Canon code actually processes it, there is some delay (there are lots of tasks processing this request, it goes through some message queues and it ends up in the MPU, probably the real-time CPU). There is no exact control over this timing; I have only measured the repeatability of the process. Each camera may require a different bias though.

Maybe the script should give the camera the closest shutter time that is actually used by the system.  For example instead of 0.1815 it should send 1/5 which is the closest camera acceptable shutter.  By doing this maybe it is possible to avoid some of these camera interpretation errors?

Again this does not seem to be a problem in the bulb mode but it is more pronounced below the 1 sec shutter speed.

Just and idea..
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

RenatoPhoto

I am wondering if the Timelapse Bulbramping script could send the exact shutter speeds to avoid some of the errors I found on the log files.  I suspect that this is the cause of flicker around the shutter speeds below 1 second as previously noted.

I tried to implement some lines of code (but unfortunately I am not a programmer) that could do this so I can compile and use it with 5D3.  But after a few tries I am not sure if I have the correct variable and I don't even know exactly where to put in the shoot.c script.

So I just note it here in the slight case that someone might be interested in implementing this:

if (bulb_shutter_valuef<=0,90000 && bulb_shutter_valuef>0,70000) bulb_shutter_valuef=0,8
if (bulb_shutter_valuef<=0,70000 && bulb_shutter_valuef>0,55000) bulb_shutter_valuef=0,6
if (bulb_shutter_valuef<=0,55000 && bulb_shutter_valuef>0,45000) bulb_shutter_valuef=0,5
if (bulb_shutter_valuef<=0,45000 && bulb_shutter_valuef>0,35000) bulb_shutter_valuef=0,4
if (bulb_shutter_valuef<=0,35000 && bulb_shutter_valuef>0,27500) bulb_shutter_valuef=0,3
if (bulb_shutter_valuef<=0,27500 && bulb_shutter_valuef>0,22500) bulb_shutter_valuef=0,25
if (bulb_shutter_valuef<=0,22500 && bulb_shutter_valuef>0,18333) bulb_shutter_valuef=0,2
if (bulb_shutter_valuef<=0,18333 && bulb_shutter_valuef>0,14583) bulb_shutter_valuef=0,166666667
if (bulb_shutter_valuef<=0,14583 && bulb_shutter_valuef>0,11250) bulb_shutter_valuef=0,125
if (bulb_shutter_valuef<=0,11250 && bulb_shutter_valuef>0,08846) bulb_shutter_valuef=0,1
if (bulb_shutter_valuef<=0,08846 && bulb_shutter_valuef>0,07179) bulb_shutter_valuef=0,076923077
if (bulb_shutter_valuef<=0,07179 && bulb_shutter_valuef>0,05833) bulb_shutter_valuef=0,066666667
if (bulb_shutter_valuef<=0,05833 && bulb_shutter_valuef>0,04500) bulb_shutter_valuef=0,05
if (bulb_shutter_valuef<=0,04500 && bulb_shutter_valuef>0,03667) bulb_shutter_valuef=0,04
if (bulb_shutter_valuef<=0,03667 && bulb_shutter_valuef>0,02917) bulb_shutter_valuef=0,033333333
if (bulb_shutter_valuef<=0,02917 && bulb_shutter_valuef>0,02250) bulb_shutter_valuef=0,025
if (bulb_shutter_valuef<=0,02250 && bulb_shutter_valuef>0,01833) bulb_shutter_valuef=0,02
if (bulb_shutter_valuef<=0,01833 && bulb_shutter_valuef>0,01458) bulb_shutter_valuef=0,016666667
if (bulb_shutter_valuef<=0,01458 && bulb_shutter_valuef>0,01125) bulb_shutter_valuef=0,0125
if (bulb_shutter_valuef<=0,01125 && bulb_shutter_valuef>0,00900) bulb_shutter_valuef=0,01
if (bulb_shutter_valuef<=0,00900 && bulb_shutter_valuef>0,00713) bulb_shutter_valuef=0,008
if (bulb_shutter_valuef<=0,00713 && bulb_shutter_valuef>0,00563) bulb_shutter_valuef=0,00625
if (bulb_shutter_valuef<=0,00563 && bulb_shutter_valuef>0,00450) bulb_shutter_valuef=0,005
if (bulb_shutter_valuef<=0,00450 && bulb_shutter_valuef>0,00356) bulb_shutter_valuef=0,004
if (bulb_shutter_valuef<=0,00356 && bulb_shutter_valuef>0,00281) bulb_shutter_valuef=0,003125
if (bulb_shutter_valuef<=0,00281 && bulb_shutter_valuef>0,00225) bulb_shutter_valuef=0,0025
if (bulb_shutter_valuef<=0,00225 && bulb_shutter_valuef>0,00178) bulb_shutter_valuef=0,002
if (bulb_shutter_valuef<=0,00178 && bulb_shutter_valuef>0,00141) bulb_shutter_valuef=0,0015625
if (bulb_shutter_valuef<=0,00141 && bulb_shutter_valuef>0,00113) bulb_shutter_valuef=0,00125
if (bulb_shutter_valuef<=0,00113 && bulb_shutter_valuef>0,00090) bulb_shutter_valuef=0,001
if (bulb_shutter_valuef<=0,00090 && bulb_shutter_valuef>0,00071) bulb_shutter_valuef=0,0008
if (bulb_shutter_valuef<=0,00071 && bulb_shutter_valuef>0,00056) bulb_shutter_valuef=0,000625
if (bulb_shutter_valuef<=0,00056 && bulb_shutter_valuef>0,00000) bulb_shutter_valuef=0,0005
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

a1ex

I don't understand what you are trying to do; ML already uses exact shutter speeds below 0.5 seconds.

RenatoPhoto

That is not what the bramp.log file shows.  So if you are correct the information about the shutter speed in the log is incorrect.  But here I present the information in the log file and the actual shutter speed shown on the RAW image in photoshop.  This information is extracted from: http://www.pululahuahostal.com/ML/BRAMP_GRAPH-5D3.xlsx

Photo   Shutter        Shutter   
Number   from Log     from Raw   Error %
4531   0,83468      0,8                 4,335
4532   0,72116      0,7                 3,022857143
4533   0,62862      0,6                 4,77
4534   0,55724      0,5                 11,448
4535   0,49851      0,5                 -0,298
4536   0,44339      0,5                 -11,322
4537   0,38657      0,333333333   15,971
4538   0,33706      0,333333333   1,118
4539   0,29125      0,333333333   -12,625
4540   0,25294      0,2                  26,47
4541   0,21916      0,333333333   -34,252
4542   0,18155      0,2                  -9,225
4543   0,15323      0,2                  -23,385
4544   0,12489      0,1                  24,89
4545   0,09794      0,1                  -2,06
4546   0,08112      0,076923077   5,456
4547   0,07063      0,066666667   5,945
4548   0,06311      0,066666667   -5,335
4549   0,05686      0,066666667   -14,71
4550   0,05055      0,05                  1,1
4551   0,04537      0,05                  -9,26
4552   0,04038      0,04                  0,95
4553   0,03626      0,033333333   8,78
4554   0,03330      0,033333333   -0,1
4555   0,03052      0,033333333   -8,44
4556   0,02781      0,033333333   -16,57
4557   0,02509      0,025                  0,36
4558   0,02285       0,025                  -8,6
4559   0,02084       0,025                  -16,64
4560   0,01902      0,02                     -4,9
4561   0,01751      0,016666667         5,06
4562   0,01647      0,016666667         -1,18
4563   0,01550      0,016666667         -7
4564   0,01459      0,016666667         -12,46
4565   0,01365      0,016666667         -18,1

Clearly the BRAMP.log file does not show exact (Canon) shutter speeds listed below beginning from 0.8:

0"8, 0"6, 0"5, 0"4, 0"3, 1/4, 1/5, 1/6, 1/8, 1/10, 1/13, 1/15, 1/20, 1/25, 1/30, 1/40, 1/50, 1/60, 1/80, 1/100, 1/125, 1/160, 1/200, 1/250, 1/320,1/400, 1/500, 1/640, 1/800, 1/1000, 1/1250, 1/1600, 1/2000, 1/2500, 1/3200, 1/4000, 1/5000, 1/6400, 1/8000
         
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

RenatoPhoto

Ok, finally uploaded the Timelapse for the 5D3 and the 7D.   All of the analysis is above.

5D3 Timelapse using BulbRamp Smooth  https://www.youtube.com/watch?v=QAaZtlnKlKg

7D Alpha 2 Timelapse BulbRamp Sunrise:  https://www.youtube.com/watch?v=Ol3YMvc3qWM
http://www.pululahuahostal.com  |  EF 300 f/4, EF 100-400 L, EF 180 L, EF-S 10-22, Samyang 14mm, Sigma 28mm EX DG, Sigma 8mm 1:3.5 EX DG, EF 50mm 1:1.8 II, EF 1.4X II, Kenko C-AF 2X

tinyenormous

Hi a1ex,  I have bramp_parse_log.py installed as well as its dependencies. I'm embarrassed to admit it, but I don't know how to properly pass an image file into it!

The best I can get is this.

ImportError: dlopen(/Library/Python/2.7/site-packages/numpy-1.8.0.dev_4600b2f_20130131-py2.7-macosx-10.8-intel.egg/numpy/core/multiarray.so, 2): no suitable image found.  Did find:
   /Library/Python/2.7/site-packages/numpy-1.8.0.dev_4600b2f_20130131-py2.7-macosx-10.8-intel.egg/numpy/core/multiarray.so: mach-o, but wrong architecture

Thank you.
FWIW I got your other deflickering script installed and running under wine. I've only used it once, but it completed its batch.