cantsin

On top of that, a larger sensor will always have a better signal/noise ratio than a smaller sensor using the same sensor tech. Which means that 100 ISO on MFT will be as noisy as 200 ISO on APS-C and 400 ISO on Full Frame etc. (And conversely: that ca. 6400 ISO is currently the upper limit for high-quality low-light images on full frame [even on the A7s if you look at the noise in unprocessed raw stills], 3200 ISO for APS-C and 1600 ISO for MFT.

Light gathering of MFT camera factually IS worse than APS-C and Full Frame because camera manufacturers senselessly push the megapixel count on the sensors, resulting in more noise and limitations at high ISOs that could be avoided with bigger sensor pixels. With today's sensor technology, the 'sweet spot' for resolution vs. noise is 42MP for full frame, 24 MP for APS-C and 12 MP for MFT. (Above that, noise increases, dynamic range decreases, and you will have a hard time finding lenses that can actually resolve the higher pixel count.)

Forget raw video editing on a run-of-the-mill laptop, with the sole exception of industrial heavy-duty laptops using desktop components (i.e. desktop CPUs & GPUs, plus a combination of system SSD and data HDD) like they're being offered as barebones by Clevo and sold as ready-to-use laptops by different smaller manufacturers. - Also make sure to have a top-of-the-line TFT display that is good enough for critical color correction/grading work. Most gamer laptops disqualify because of their low-resolution TN displays.

It bugs me that the word "banding" is wrongly used in forums everywhere. "Banding" doesn't refer to shutter artifacts, but to artifacts caused by a lack of color depth (such as: color stepping in blue sky gradients).

MagicLantern gives us some insight on that laser-focus, too. That Canon has switched the internals (firmware/operating system, likely also electronics) of the EOS-M series from EOS xxxD to PowerShot, is genius from a cost-saving business perspective. But terrible from a technological point of view.

If you want natural/unprocessed/organic look of images and deep colors, go for the BM Pocket at all means. The GH5, despite its many virtues (especially as a high-quality, compact, ready-to-shoot camera with optical sensor stabilization, good internal sound recording and no need to always have IR cut filters in front of the lens), has a much more processed image. But you can best judge this by yourself by comparing videos on Vimeo: https://vimeo.com/groups/blackmagicpocketcinema/videos https://vimeo.com/groups/389055/

@mercer, there are c-mount to d-mount adapters but not the other way around. (Which would be as difficult or even impossible as adapting rangefinder or mirrorless lenses to SLRs.) - The subsequent EOS M models don't support MagicLantern. Except perhaps for the M2, they aren't likely to ever run it since they're not based on EOS, but on lower-end PowerShot firmware. The EOS-M almost never freezes or stops if you restrict the raw video resolution to 720p. EDIT: I found this: http://www.calkovsky.com/d-mount-to-c-mount-lens-adapter-convert-d-mount-to-c-mount/ ...but it will effectively turn adapted d-mount lenses into macro lenses since d-mount has a flange distance of 12.29 mm whereas c-mount has 17.526mm. (In other words, a better-made adapter would have to recess the d-mount lens 5.236mm into the camera body.)

Chiming in with two short videos shot with Canon EOS-M raw crop mode and adapted c-mount/cine lenses under my belt: In MLV raw crop mode, the EOS-M's effective sensor size is equivalent to Normal (/Double) 8mm film (4*3mm) when filming in 960*720p and Super 8 (5.5*4mm) when filming with 1280 pixels horizontal resolution. So the best-fit lenses are those from old Super 8 cameras - with the caveat that only few Super 8 cameras had interchangeable lenses. The above video was shot with a Schneider Variogon 6-66mm/f1.8, an M mount lens originally designed for the Leica Leicina Special (one of the best Super 8 cameras ever built). I used it with a cheap Leica M-to-EOS-M adapter but am upgrading to a Novoflex since the lens isn't parfocal with the cheap adapter. (It is on the Leicina, like practically all Super 8 camera zooms.) The c-mount options include Angenieux zooms built for Beaulieu Super 8 cameras and a 7.5-75mm/1.8 Fujinon built for the Single-8 Fujica ZC1000 (another great camera) - however, many if not most of them will require mechanical modification to fit EOS-M adapters. The bulk of Normal 8 camera lenses (which include Taylor-Hobson/Cooke, Kern Switar, Cine Nikkor, SOM Berthiot and many more) are still out of reach because they're made for d-mount, with smaller diameter and shorter flange distance than c-mount, and there are no adapters for EOS-M (yet...). Oh yeah, and the EOS-M doesn't require h264 recording parallel to raw in resolutions under 1920p.

I should have been more specific: The Makefile doesn't work with Linux (because of a lack of specified targets), and the Windows version doesn't work because of a weird dependency of the FUSE replacement software on an outmoded, uninstallable version of the Microsoft Visual C++ runtime. This looks like software that should run on Linux and Windows in theory, but is being written by developers who neither use these two systems, nor run tests on them. EDIT: My bad. Made a mistake compiling, pulling the fresh repository via git and then running make solved the issue.

I tried, but it doesn't work....

What you're looking for is a filter that simulates the jitter and gate weave of analog film transport, not a handheld camera shake simulator. Red Giant's Misfire plugin (which is part of the Magic Bullet suite) does exactly that.

Here's another video shot with the EOS-M with MagicLantern in 960x720p raw using the sensor crop mode and an adapted Schneider Optivaron 6-66mm/f1.8 lens originally designed for Leicina Special Super 8 camera. (In crop mode, the effective sensor size is practically equivalent to 8mm film.) Everything shot handheld with a small Braun chestpod originally designed for Nizo Super 8 cameras.

Note that Schneider's Arriflex mount lenses don't officially cover Super 16 (the equivalent of a 1" sensor), but only 4:3 16mm film (the equivalent of a 2/3" sensor). From 25mm onwards, lenses will cover the sensor without crass vignetting, but the left and right corners will be outside the official image circle and therefore blurry.

Used chroma smoothing via raw2cdng.

Here's a video shot with the Canon EOS-M in Magic Lantern raw sensor crop mode and an adapted Fujinon 14-84mm/f1.6 c-mount zoom lens. Graded and edited in Resolve, with grain added via Filmconvert. The inspiration were Jonas Mekas' 1960s experimental diary films shot handheld with a Bolex H16 - so, no apologies for the shaky handcamera, jump cuts, out-of-sync sound etc:

Neat Video. Essential.

Isn't the Angenieux only covering 16mm 4:3 and vignetting on the BM Micro Cinema's 1" sensor? Did you fully get around this limitation with the 120% enlargement in post, or did you only use a limited zoom range of the camera? (Since 16mm zooms usually vignette at wide angle settings, but cover larger sensors when zoomed in.)

Rode VideoMicro. Unbeatable for the price (ca. $60), very small/light/portable, runs without batteries, decent audio quality, significant improvement over internal mics.

Point taken with the EVF, my bad - had confused the A99 with the Sony's translucent mirror SLRs in that detail. However, the Minolta manual focus lenses have a different mount (MD) that can't be adapted to A mount without losing infinity focus. So no luck for your 50mm/f1.2.

Why not just use a Sony EA3 adapter on an A7 camera? In the end, the A99 is just an A7RII with a mirror box that will be useless for video. Coupled with the EA3, the A7RII will effectively be an A99, but with the added benefit of a built-in EVF. - Btw., an upside of A-mount lenses (including Minoltas) is that they're easily to find for a few bucks in charity stores. But just like with Canon's EF mount, the lenses don't have aperture rings, and only tiny, fast-travelling focus rings that aren't really suitable for manual focus pulling.

The color shifts are most likely cause by the fact that browsers like Firefox and Safari support color management (i.e. the system-wide ICC monitor profile) while Chrome lacks color management support.

Optical flow has been in Resolve since version 10.

Because m43 lenses aren't adaptable and therefore can't be migrated to other camera systems - while you can adapt a full frame Nikon lens to almost any camera system currently in use; when using a Speed Booster/focal reducer, even with higher apertures/lens speeds. On top of that, electronic m43 system lenses are optimized for being lightweight and small and therefore involve a lot of optical design compromises - above all, firmware-/software-based geometry correction and devignetting. (I.e. your Panasonic or Olympus 12mm, 14mm, 17mm or even 20mm m43 lens actually happens to be a slight fisheye lens that relies on in-camera photoshopping - whereas all full frame DSLR lenses on the market are optically corrected.) On top of that, m43 lenses involve compromises in handling and built quality: plastic instead of metal, focus-by-wire instead of a mechanical focus ring, no aperture ring. MFT lenses without these design compromises, such as the Voigtlanders and Veydras, typically cost as much as full frame lenses.

The Sony A7 doesn't have Slog or advanced video functions either. Better wait for the likely future releases of an A9R and A9s.

Also: a largely unprocessed image without in-camera, baked-in sharpening, denoising and lens geometry correction. If there is no raw recording, there must be late log profile, and at least 10bit (better 12bit) color depth. Sound recording is off-camera, so in-camera sound recording might be missing or just rudimentary for reference tracks. The sensor doesn't need to be s35mm though, there have been.s16mm cinema cameras, too, both analog and digital. There must be support for rigging, I.e. external power and external field monitoring.