Is the EOS-M *THE* Digital Super-8 Camera?

Started by lintel, January 28, 2018, 01:37:40 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

lintel

I've been playing half-heartedly with ML on my EOS-M for a while but in general lost interest due to annoyances such as the focus pixels, low resolution and workflow. However, inspired by recently released footage from Kodak's new Super-8 camera and fond memories of playing with real Super-8 footage in art school, I decided to mess around with the EOS-M again to see how close I could get to the Super-8 aesthetic. Turns out I surprised myself!

I've only shot a few early tests so far but the following video is, I think, very usable as a digital Super-8 stand-in. The basic specs are as follows:

4:3 ratio (like Super-8)
1440 x 1078 resolution (plenty for Super-8)
18fps (same as consumer Super-8)
12-bit Lossless RAW (continuous!)
5x zoom (haven't accurately calculated crop factor/sensor size in relation to Super-8 yet - would appreciate help with this)
3X3 Crop mode (Experimental build: magiclantern-crop_rec_4k.2017Dec19.EOSM202)
Also- no focus pixels! (I have no idea why this is. They reappear if you go up to 14bit lossless or use any of the standard 10/12/14-bit modes).

Below are two versions of my initial tests - the first is with FilmConvert film emulation applied, the second is without FilmConvert, just some basic contrast and saturation tweaks. Sorry the grade isn't very good I'm getting used to both the EOSM DNG's and FCPX's new grading tools.

FilmConvert:
https://vimeo.com/253014693/badd381eb4
(how does one embed on here btw?)

No FilmConvert:
https://vimeo.com/253014620

My post workflow is very simple: convert .MLV files to DNG's with MlRawViewer and then drop them straight into FCPX. Do a bit of colour correction with an adjustment layer if needed, then export clip as ProRes 4444 XQ. Bring it back into FCPX and conform the 18fps to the fps of the timeline (I always use a 25p timeline and *think* this requires slowing the 18fps clip to 72% but I'm not confident my maths are accurate - it certainly looks close to normal speed to me though).

I'm not very technical so you won't find much explanation of why it's working from me. I'd appreciate contributions to what's going on here so I understand it better and other can replicate if they want to. In particular I'd like to figure out the effective sensor size I'm using, and also why there are no pink dots.

The things that excite me about it as a viable Digital Super-8 camera are:
1. The shutter speed - only Magic Lantern allows that really, and it really helps give that authentic Super-8 feel.
2. The 12bit colour space and RAW grain makes the footage film-like and organic.

3. The fact you can adapt C-mount lenses to the EOS-M. The lens I used for these tests is just the 15-45mm EF-M kit lens in manual focus mode. However I have a Cosmicar 6mm f/1.2 on its way to me as we speak and if I can get it to infinity focus I think I'll have a 'normal' lens (again, I need to figure out the imaging area I'm working with).

4. Shooting 4:3 and 18fps, like Super-8, allows continuous shooting.

Things I don't like:
1. Live view is not perfect but using the info button you can jump between Canon's 5x zoom to focus and ML's rather choppy live view for framing.
2. I'm not seeing horrible rolling shutter but it's no Digital Bolex so that does give it away as digital footage somewhat.

I hope some others start playing around with these settings. With the 16mm Digital Bolex discontinued and Kodak's new Super-8 camera all the rage, there's a space for a small sensor digital cinema camera to get some love.

dfort

Quote from: lintel on January 28, 2018, 01:37:40 PM
(how does one embed on here btw?)

Like this:

[vimeo]https://vimeo.com/253014693[/vimeo]

FilmConvert:


No FilmConvert:


It is pretty cool that C-mount lenses adapt so easily to the EOSM.

To mimic 8mm I'd suggest using the "Movie Crop Mode" instead of the zoom mode. The area of the sensor used in this mode is roughly 1/3 of the full sensor. Why? Because instead of binning/skipping every 3 pixels in order to reduce the resolution of the full sensor to video resolution it uses all of the pixels but in a smaller area of the sensor. That would translate to an image size that is about 7.5mm wide - give or take. That's something between Super-8's 5.79mm and 16mm's 10.26mm. (Regular 8mm is 4.5mm and Super-16mm is 12.52mm.) If you want to get closer to 8mm try reducing the image size to 1280x960 for a 4:3 aspect ratio and change the Preview to Real-time for a better user experience. For accurate framing I used to put tape on the LCD screen thought things have gotten better lately.

Why not zoom mode? You don't really need the extra horizontal resolution, preview mode doesn't work as well and it is harder to get into this mode on the EOSM because the magnifying glass button is on one of the Canon screens, not the ML menu as in Movie Crop Mode. By the way zoom mode is often called 5x zoom because it normally shows a 5x magnification of the image on the Live View screen. You can pan the focus area around the image area so you're not quite sure if you are centered on the sensor when you shoot and Real-time previewing is usually not very accurate so it is a choice between accurate framing and slow refresh rates or inaccurate framing and Real-time Live View.

However, if you want the largest, best quality image size then by all means use zoom mode. The EOSM can now go all the way to 2520x1080 in zoom mode.