An exposure is when you expose the sensor to light by opening the shutter. That only happens once with this technique. Or if you prefer another definition of exposure, it's how much light is allowed to fall on the sensor. There's still only one level of exposure with this technique. What there's two of is gain settings. Gain is not part of exposure. Of course there are HDR techniques that use two exposures.

It's bad enough that we already have two definitions of exposure. Let's not make up a third one. :)

BMCC sensor: http://www.scmos.com/files/high/scmos_white_paper_8mb.pdf

BMPCC sensor: http://www.fairchildimaging.com/files/data_sheet_cis_1910f_0.pdf

The two Blackmagic sensors do read-out the same way. See page 7 of the white paper.

OK. Lets not argue about it. It depends which way you want to look it. Basically each of the two analog amplifiers are programmed to take a user-defined ISO 'exposure' (e.g. ISO 100 / ISO 1600).

build link for 5dmkiii: https://drive.google.com/folderview?id=0B7QlH_BH2m32aWFxbm9DNkliS0E&usp=sharing

Is it true that BMCC does this?

@Oedipax: Thread and download link here: http://www.magiclantern.fm/forum/index.php?topic=7139

The aliasing seems to be very bad on the video above. There also seems to be a loss in resolution.

@driftwood Just a quick clarification - the "radioactive look" has nothing to do with HDR, bracketed, exposures, etc. Nothing at all.

It is a purely a function of certain tonemapping algorithms and can be achieved on single exposure LDR sources both quickly and easily.

As someone who frequently shoots well over 600,000 timelapse exposures in a year (with a large percentage of them bracketed HDR) I cannot emphasize enough how completely the look is dependent on the approach to post-processing and specifically to the tonemapping algorithm and settings used.

So, once again - the "radioactive look" has NOTHING to do with HDR. It has to do with people doing garrish things with tonemapping. Okay, done now. :)

@driftwood Looks like it's got definite potential - it's coming along quicker than I might have expected. :) Thanks for sharing the very useful test clip.

Great lill' 5DMKIII ML RAW video here. Especially the last shot.

Just finished up an interview shot on both the 5DM3 (CU) and the 50D (Medium). The cameras worked well together, but looking back I would have made the 50D the CU camera in order to avoid the moire. However, I just updated the 50D to the July 11th build and tested the crop mode at 1920 1:85 and it recorded continuously on my 32GB KomputerBay 1000X CF for 5gb before I stopped it! This is great! I'll be sure to use this crop mode next time when I shoot alongside the 5DM3.

[vimeo]

@flablo Whatever does it for you really. Looking at some of the footage coming out of the BM Pocket it does look rather splendid and its nice n cheap. Just dunno how long it would take to get one. Ive been messing with the 5D3 ML RAW and its a remarkable hack. However, the price of the cam plus lenses if you don't already own them is going to be steep. Personally, I wouldn't sell / wait and see for a few months and judge reactions from consumers. I really wanna do my own comparative tests.

Shot RAW on the 5DM3 and 50D using the Magic Lantern Hack. The 5DM3 was shot at 1920x1038 (1.85). The 50D was shot at 1920x1038 resolution in cropped mode to match the 5D. A pair of 64GB 1000X KomputerBay CF cards were used for the shoot. The ratio of 5D/50D is approximately 60/40. Can you tell which shots are which?

The RAW files were converted to dng sequences with Rawanizer ver 0.5.5. DNxHD proxies were used to edit the footage in Adobe Premiere CC. Once edited, the sequence was brought into Adobe After Effects CC via Dynamic Link and the proxies were replaced with the RAW footage. First light color correction was done in ACR, and additional corrections were performed in AE. All compositing was performed within AE. Gorilla Grain was added to the final output.

Externally.