T2i Magic Lantern for 550D in progress!

Hi, here some new tests I did with the new build.
Every shot is about 20 seconds on a static scene with ISO 1250 and CineStyle.

FW Standard
View attachment 34431

CBR 0,1
View attachment 34432

CBR 2,0
View attachment 34433

CBR 3,0
View attachment 34434

Qscale 16
View attachment 34435

Qscale 8
View attachment 34436

Qscale 0
View attachment 34437

Qscale -8
View attachment 34438

Qscale -10
View attachment 34439

Qscale -12
View attachment 34440

Qscale -16
View attachment 34441


Best regards,
Dorian

Have you ever experienced the recording being stopped with FW default + audio with these peaks?
 
I confirm this bug was not fixed, see here: http://groups.google.com/group/ml-d.../a152b31eb2ca4144?show_docid=a152b31eb2ca4144

With yesterday's build, users of 30fps will experience bitrates higher than normal :)

Today's build should be OK, but testing is needed to make sure there are no more bugs. At least CBR 0.1x works now :)

I was referring to another bug, before I had the problem with peaks (and recording stopping), then with newer fws I had that cbr1.0/fw default had lower bitrates than without ML (same scene on tripod 44avg/52peak vs 38avg/41peak covering the lens in bright daylight scene). Now it's exactly the same. I've used only 24p though.
 
Have you ever experienced the recording being stopped with FW default + audio with these peaks?
Hmm, actually I experienced some recording stops with this settings. But these are not many and a friend with the same camera (without ml) had the same I guess (he has a cheap 16gb card though).

Can anybody else check some MOVs with the bit rate viewer and show the graphs?

Best regards,
Dorian
 
The peaks (and lack of them in my tests) were because I used BitrateViewer in second-based mode, and Dorian used it in frame-based mode.
 
I have a question about sdhc cards with magic lantern loaded and bootable: Do I need to reformat these cards before using them with another camera? I will be using my T2i and GoPro on a bike trip this summer. I won't have access to a computer regularly so ideally I will be swapping my two formatted cards back and forth between the two cameras without reformatting every time. Is this sound? or will it potentially brick my GoPro?
 
I have a question about sdhc cards with magic lantern loaded and bootable: Do I need to reformat these cards before using them with another camera? I will be using my T2i and GoPro on a bike trip this summer. I won't have access to a computer regularly so ideally I will be swapping my two formatted cards back and forth between the two cameras without reformatting every time. Is this sound? or will it potentially brick my GoPro?

nope, you'll be fine swapping as your gopro isnt LOOKING for anything on the card to load. Its just using the card for file placement.

Remember the first step in loading ML is to load that custom firmware, that is where the change initially starts.. By default the t2i doesnt look on the card either for anything to load and just uses the card to place files. :)
 
nope, you'll be fine swapping as your gopro isnt LOOKING for anything on the card to load. Its just using the card for file placement.

Remember the first step in loading ML is to load that custom firmware, that is where the change initially starts.. By default the t2i doesnt look on the card either for anything to load and just uses the card to place files. :)

Well said, thank you for the clarification.
 
Hi Alex.
I am holding a 600d, I have not found many differences between the 550D and 600D but the display tilt and 3X zoomrec 10X ... Whereas the day cameras are virtually identical believe that we can implement the reczoom also T2i? I would add an entry on DollyZoom ML and include this function so far on T2i can not be completed.
... would also be very interesting for those who like me have a 30mm f1.4 that works well at night and I pleased to be able to use it to 3X (taking the f1.4) would become a 90mm f 1.4 ... I did some testing and the quality is acceptable to 3X.
I hope you too will experience the DollyZoom and that the idea is interesting for you too!

A example of a VertigoDollyZoom in the film La Haine : http://www.youtube.com/watch?v=iv41W6iyyGs
or http://www.youtube.com/watch?v=LutRaK8YmA0&feature=related
 
... would also be very interesting for those who like me have a 30mm f1.4 that works well at night and I pleased to be able to use it to 3X (taking the f1.4) would become a 90mm f 1.4 ... I did some testing and the quality is acceptable to 3X.

You do know that that your so called Dolly Zoom effect is a fine example of why a 30mm with 3X Zoom is not a 90mm. The way the light is effected by the glass is different. Your DOF and FOV will be different. Will it work? Yes. Will it be off putting to the viewer? Maybe.
 
Isn't the idea of a Dolly Zoom to be "off putting" in the first place?

Still not sure this would work the way he is describing. But hey, having the zoom feature on the T2i at all would be pretty cool, but I'm not holding my breath.
 
You do know that that your so called Dolly Zoom effect is a fine example of why a 30mm with 3X Zoom is not a 90mm. The way the light is effected by the glass is different. Your DOF and FOV will be different. Will it work? Yes. Will it be off putting to the viewer? Maybe.

I am interested to know more about the characteristics of the 3x zoom effect. Could you explain why the FOV on a 30mm with 3x would be different to a true 90mm? I see that DOF would be much shallower on a true 90mm at f/1.4, because what you are doing is reducing the sensor size in effect.

How about exposure? How much light would fall on the sensor with the 3x 30mm vs a 90mm both at f/1.4?
 
with a 30mm f/1.4 and 3x digital zoom, you get a 90mm f/4.2 in terms of FoV and DoF
http://www.similaar.com/foto/dof/dof.php?f_x=1.6&f_r=1.5&f_mm=90&f_f=4.2&f_d=2
http://www.similaar.com/foto/dof/dof.php?f_x=4.8&f_r=1.5&f_mm=30&f_f=1.4&f_d=2

in terms of exposure, it's a 90mm f/1.4

BUT in the future you might want to use lower ISO values

in terms of light input, as long as we're using line skipping, the number of photosites actually used (and their cumulative area) may not be changed by the 3x zoom; if that is so, that lens is going to be like a 90mm f/1.4 in terms of light input

if we were using a bicubic algorithm or something like that, you'd have less light hitting the used parts of the sensor, and so more noise; or, conversely, you'd have to reduce your ISO to keep the noise constant, so it would become a 90mm with something between f/1.4 and f/4.2 also in terms of light input (it depends on how linear the noise canceling is; I guess it would be closer to f/4.2)


what this 3x crop mode would do for me is make my 300mm f/4 into a 900mm that is absolutely perfect for a full moon shot at 2.35:1
ET here I go!! :)
 
Last edited:
with a 30mm f/1.4 and 3x digital zoom, you get a 90mm f/4.2 in terms of FoV and DoF
http://www.similaar.com/foto/dof/dof.php?f_x=1.6&f_r=1.5&f_mm=90&f_f=4.2&f_d=2
http://www.similaar.com/foto/dof/dof.php?f_x=4.8&f_r=1.5&f_mm=30&f_f=1.4&f_d=2

in terms of exposure, it's a 90mm f/1.4

BUT in the future you might want to use lower ISO values

in terms of light input, as long as we're using line skipping, the number of photosites actually used (and their cumulative area) may not be changed by the 3x zoom; if that is so, that lens is going to be like a 90mm f/1.4 in terms of light input

if we were using a bicubic algorithm or something like that, you'd have less light hitting the used parts of the sensor, and so more noise; or, conversely, you'd have to reduce your ISO to keep the noise constant, so it would become a 90mm with something between f/1.4 and f/4.2 also in terms of light input (it depends on how linear the noise canceling is; I guess it would be closer to f/4.2)


what this 3x crop mode would do for me is make my 300mm f/4 into a 900mm that is absolutely perfect for a full moon shot at 2.35:1
ET here I go!! :)

Thanks for that. This sounds about right to me.

Given that dividing the width of the sensor by 3 gives 1728, it's clear there is some interpolation in the works. As far as I have heard, aliasing and moire are reduced in the zoom range, which would perhaps suggest it's more interpolated than line skipped, no?

There has been a suggestion that the non-zoomed image from a t2i, while recorded as 1080p, is captured from a smaller array of pixels than 1920x1080. If this is so, doesn't that complicate the comparison of possible light gathering power between 3x and normal? Maybe they are similar. I would love to see some testing.

In any case, I can't wait til the price drops a bit and ML matures on the t3i, because when this happens, I will be trading up in a heartbeat.
 
Also, doesn't the crop mode jump between a choice of zooms? Since it just switching which pixels are being recorded from on the sensor, its in no way a "smooth zoom" like Mirko wanted for a Dolly Zoom effect anyway.

As far as the less moire/aliasing when zoomed, as I understand it the moire artifacts are lessened because we aren't skipping entire rows of sensor data in between pixels anymore (since they are closer together).
 
Sorry for being the dude to ask a questions that's probably answered somewhere in the last 131 pages but is anybody using ML with an external monitor? Is it usable at this point?
 
Also, doesn't the crop mode jump between a choice of zooms? Since it just switching which pixels are being recorded from on the sensor, its in no way a "smooth zoom" like Mirko wanted for a Dolly Zoom effect anyway.

As far as the less moire/aliasing when zoomed, as I understand it the moire artifacts are lessened because we aren't skipping entire rows of sensor data in between pixels anymore (since they are closer together).

I got the impression it was a gradual zoom. Check this out, from 02.50 onwards:

http://vimeo.com/20420054
 
Back
Top