Leaderboard

To expand on it, let's take 24 frames or one second at film rate. With film you have 24 full images, motion blur is kept in each image, not smeared netween them. With ML 5D raw, Alexa Pro Res, Red Raw, Blackmagic cameras etc, you have 24 distinct images, individually compressed either losslessly or mildly lossily. Motion blur is kept within each image. With AVCHD and other Long GOP CODECs, the I-frame image is divided into pools of pixels for analysis across time. only some frames are whole, the I frames, the rest refer the different pixel pools to the I frame and usually to a frame in the middle of the group too. In order to remove data the codec only moves what it has to. Your one second may have only two complete images, the rest are created as best can be from the other frames, with huge changes causing pixellation and blocking due to so little data headroom. On top of this, the red-channel resolution is quartered, and the blue channel halved in 4:2:0, so the spatial resolution for movement is cut down, as well as the temporal resolution. In short, the plastic movement you get from many implementations of such long GOP CODECs reflects the methods use in compression. Motion blur is also smudged and blurred by chroma sub-sampling (4:2:0:) and by interframe compression.

The Alexa has a really fast sensor readout, 2ms... motion cadence on that is of course fine as it is on all of ARRI's cameras, so that probably excludes fast readout as being the issue. Maybe some cameras implement the shutter timings differently to others? Say if you are in 1/50 or 1/40 or 180deg shutter the sensor might still remain in 60fps and it will just expose for longer and then the image processor converts it from 60p to 24p and then the codec maybe fudges things further with trickery like inter-frame compression and inter-frame noise reduction, whereas on a simpler camera like the Digital Bolex there is none of that shit and basically a global shutter sensor and entire image pipeline optimised for 24fps at 180 degree shutter. Maybe some of our modern mirrorless camera sensors have a true 24fps mode and others only have 60fps mode, process the 24p from that. Maybe some codecs are truly awful at motion... On the 50D it didn't have a video mode at release if I remember correctly? Magic Lantern enabled it via the raw video hack, but there's no H.264 mode is there? I could understand why some of the older H.264 low bitrate codecs hurt motion a lot, Cinema DNG and Magic Lantern were like going from digital video to film. The leap was that big. I should do some tests in Cinema DNG on my Sigma Fp-L as it is a direct sensor feed, uncompressed RAW and no pixel binning in the 1.3x crop mode. I bet that looks pretty good at 180 degree shutter in 24p.

Ah, my favorite topic (and it comes up every couple of years). Hopefully we can have this discussion without the usual accusations. I've done a lot of personal experimentation in this area because most cameras looked off to me. Where I've landed is that I don't think that global shutter or CCD sensors is a surefire recipe for pleasing motion. I spent a lot of time and money figuring that one out. My list of favorite motion capture cameras is a motley crew and includes the Red One MX (but oddly, not the DSMC1 line), the EOS-M in 1:1 crop, and believe it or not, the hacked GH1.

Agreed completely on your assessment of the image characteristics. I truly appreciate the performance of classic anamorphic adapters and am happy that they are coming down in price. I just picked up a baby Kowa Anamorphic 8 2X for cheap and it's a jewel. A few examples of what I consider ideal performance below. First three are the ZIA 1.5x with an older Leica 50 Elmar collapsible, which is a tiny and wonderful combo that just covers full frame 3:2. Second is the baby hypergonar 1.75x with a Voigtlander 75 2.5, which also covers full frame but is a bit more muted (although this adapter can be fickle against the light sometimes).

I've been thinking about global shutter for a specific project, after being unsatisfied with test shoots using my S5. There are several to choose from now, but you're totally right that it's surprisingly difficult to find quality reviews or comparisons, particularly older models like the F55 in a modern context. If I move forward with it, I'll probably end up with a Komodo--but I've struggled to find direct comparisons between the Komodo, F55, Ursa Mini 4K, and A9III, not to mention comparisons with IMX410 cameras that I'm already familiar with.

I have one and it really is a good camera when it comes to the image quality. I love the image it makes. It's brought my own filmmaking to another level. It's not a run and gun camera though. More of like using a 16mm film camera. Nothing automatic. Manual everything. The dream of an all-in-one isn't quite true with it though. You need to rig it out with an EVF. (Although I don't have and have never used the shark fin shaped viewer with the mirror so maybe that's okay.) I used to see them come up on eBay a few times a year but now they rarely do. And people are asking US$11000 with accessories. (I also notice that nobody is bidding on them at that price and the listing ends without any bids, then gets relisted.) I occasionally see them listed for about US$5000 but they don't last long at that price. They're a good camera, worth $5000 but not worth $10000. I think that it's no longer such a unique camera image-wise and you can get a close enough image with a used Alexa or Blackmagic Production camera 4K and be making your movies. (Still if one came up and you could afford it, get it.) If you're handy you could make your own using the same sensor. Check out this project: https://github.com/lafauxbolex https://www.youtube.com/@LaFauxBolexDigitalCinemaCamera/videos

Are you talking about the US or Venezuela? 🙂

I have the 35/50/75 Saturn in E mount and haven't experienced any focusing issues on the A1, and of course that's with no Desqueeze support. It does have a bit better EVF than the Panasonic cameras, but I haven't found the EVF to be a night and day difference. I own the 35 & 50 Saturn in L mount and have had great results on the S1RII. The 35 isn't quite as well corrected as the longer lenses, so it has a bit of character, especially against the light. Still, none of the Sirui lenses can rival the results I get with my favorite anamorphic adapters, but there is something to be said for simplicity and reliability. Also - I'm not sure if the S5II was included, but the recent update for the S1II/S1RII now provides 1.6x desqueeze. Still attached from the 35 Saturn on S1RII.

Motion cadence is a phrase I use to make clients think I'm a pro.

I agree with you that the term cadence is somewhat confusing: In fact, I would I'd dare say that most of the "cadence" problems do not exist. They are simply a question of footage not properly shot or encoded. Many, many videos on the web use fast shutter speeds in sunlight to compensate the lack of an ND filter, causing an akward cadence. Some others use a very slow shutter speed in low light to compensate the lack of proper lighting, causing an akward cadence. Some in camera codecs record "fake" progressive framerates that are actually interlaced or the footage is interlaced-deinterlaced in the NLE to fit a certain standard, causing an akward cadence. Some other times, 23.976 footage is edited and encoded as 24p, which is not exactly the same. And finally, some other times it's the players fault (though I don't own an Apple TV) or the encoded stream, or any of the absurd TV "image enhacements" of any recent TV set. Under normal circumstances, any professional camera recording true progressive footage at a proper 180º shutter angle should yield that cinematic motion. And I can include DSLRs and midrange cameras, too. You don't see those "cadence" problems when you go to the movies, even though you see movies recorded (not filmed) with a mix of Alexas, Reds, C300s, C500s and even 5Ds. Because the footage is properly shot, conformed and handled throughout the whole workflow.