kye

It's not about rules, it's about perception. We're at a funny time in history. Anyone can shoot in public with a phone and no-one bothers them. Governments and private residences can setup permanent security cameras that record people in public without their consent. It's legal in most public places (here in Australia anyway) to record video. Most private places such as museums and galleries and amusement parks and events allow photography and videography for private use. The only thing that really isn't allowed is professional shooting without a permit. Unfortunately, the way that people tell the difference between the two is by the size of the camera. If I went to a park with a bunch of other parents and we all stood in a line recording our kids, everyone with an iPhone and me with an FS5, I'm going to be seen differently. If I go to a museum and film my kids running around with a BMPCC6K and a shotgun microphone I'm going to get interrogated by security. Guerilla film-making is a phrase to indicate that you are shooting without permission. It doesn't mean that you NEED permission, it just means you don't have it - just like all the mums with smartphones don't have it either. It means that fitting in and not getting noticed matters. It doesn't mean you're doing anything wrong. If someone calls security and I have to talk with them and convince them that I'm not doing anything wrong and they walk away, that's still failure to me because I don't want to have that happen in the first place. It doesn't matter that I'm not doing the wrong thing. To give you an idea about how poorly venues are able to distinguish between pros and amateurs, I went to a temple in Bangkok and there was a sign at the entry. 8mm film cameras are ok, 16mm film cameras are not. That was 2019. I hoped that no-one would think my GH5 was a 16mm film camera.

The patent system is broken. I remember years ago seeing enough examples that were beyond ridiculous that I now just think of it as yet another system that is dysfunctional and if you have enough money you can basically do anything. I just want lots of camera manufacturers to start making smaller cameras so that it gives us solo hand-held guerrilla-shooters more options 🙂

This absolutely blows my mind and makes all kind of sense. I've made a new thread because I think it needs to be highlighted more than in here, where it'll get buried quickly. So, the GH6 could have that sensor from the UMP12K, retain the MFT mount, and just use the middle 8K part as it is MFT sensor sized? That would be incredible. Could that be mounted on an IBIS mechanism? I don't really know how those things work. In theory then, assuming it kept the GH5 ethos, it could offer 8K downsampled to any resolution you want. The UMP12K can do 8k120 and 8k160 in 2.4:1 aspect ratio, so with downsampling it could offer 120p in any resolution you wanted without cropping. That would be something to put the GH6 on the map!

I am completely blown away by this, but it's great news. In another thread, @sanveer has shared a link to Imatest who have shown that lens design might be the limiting factor to the DR of a setup, not the sensor. This is great news, especially for those of us who aren't shooting on the most pristine modern lenses, and don't want to. It means that we don't need to keep buying cameras with higher and higher dynamic range specifications. I have actually been chasing lenses that have slightly less overall contrast so that the veiling flare pulls up shadows, giving the sensor a bit more light to digitise and getting it a bit higher compared to the noise floor. This also makes the image much more like film, which has a nice shadow roll-off and looks more organic. Lots to unpack in this one.

Wonderful! More resolution wouldn't improve those images, quite the opposite I think.

you didn't need the word "res" in there... "Though at 10k maybe not." then it works either way 🙂 Heavily.

A recent Canon patent shows a RED Komodo style body with huge specifications. https://ymcinema.com/2022/02/11/canon-develops-high-end-boxy-cinema-camera/ It's an interesting glimpse into Canons thinking. Firstly, it's a dead-ringer for the Komodo, including the screen on the top. It would be great if they'll use this as a screen or if it'll just be for controls, but that's potentially promising for building a small self-contained rig. But the specs show where Canon are prepared to go. 10K or more resolution 120fps or more sensors up to 56x42 (a 4:3 with crop factor of 0.64) or even 186x56 (which is a very wide aspect ratio, but the horizontal crop factor is 0.19 !!!!!!) Now, this is a patent, and it makes sense for Canon to design a body that is future proof so they can get economies of scale and standardise on accessories etc, so these are just the upper limits of what they might actually do, but they're essentially betting that these things are likely enough that it's worth designing them in. It's time to stop thinking "more is better" and start asking "how much is optimal", because Canon will forever be your pusher..

Sorry to hear about your experience - hopefully someone on here can help you with some info. Some of the ecosystems for these cinema cameras can be pretty difficult to navigate sometimes. When RED users talk to each other it seems like they're speaking a foreign language!

When I was in the market for my GH5 I was seriously considering both the A73 and an Olympus camera. I can't remember which Olympus model it was, but it was the direct competitor to the GH5. The equation basically boiled down to this: GH5 - poor AF, great codecs including 10-bit, very good stabilisation Oly - limited to 8-bit, excellent stabilisation (better than the GH5) A73 - great AF, limited to 8-bit codecs I don't remember if the Oly had PDAF - I don't think I even knew that at the time. Later on I learned that they had it (thanks to @Dave Maze on YT) and it was one of the best kept secrets - no-one was talking about it. Anyway, I realised that it came down to AF vs stabilisation vs codecs. I used one of my existing camera setups to try to learn manual focusing, and discovered that not only was it much simpler than I thought, but I actually preferred the experience (not silently screaming in my head as I watched the camera screw up my shot) and most importantly I learned that I preferred the more human feel of MF - including the mistakes and imperfections. So that ruled out the A73, because AF was it's main feature, and left the Olympus and the GH5. I eventually decided that the 10-bit internal and higher-quality codecs were more important to me than the edge on stabilisation that the Oly had, and went with the GH5. I've often wondered why there was no love for Olympus as the quality of the footage was there (I watched a LOT of it while I was evaluating cameras) but I think it was more that no-one knew about it. The GH5 was famous (10-bit! 400Mbps! ALL-I! etc) and infamous ("Panasonic AF has ruined my life!" "I want to kill myself!!" etc), but no-one even mentioned Olympus - not for their strengths or weaknesses.

Noam Kroll just posted a new blog post comparing Prores vs h265 on the iPhone which makes interesting reading. Some aspects of the comparison are specific to the iPhone, but others apply more widely. https://noamkroll.com/filmic-pro-log-vs-prores-sample-images-test-results/ Incidentally, Noam is an example of someone who prefers shooting high-quality HD rather than lower-quality 4K or more because of the associated aesthetic, as well as practical considerations. His blog is excellent and contains lots of articles discussing various aspects of film-making. In todays internet landscape I find that he's a rare glimpse of working film-makers who are balancing aesthetic and practicalities to get the overall best results from a film-making perspective.

Great camera test - I'm a big fan of testing stuff by taking video of my family and editing it up into a finished video so you can keep it as a memory. Colour and images look really good. I also use crop modes on my setup (GH5) which can be a little softer due to the limited resolution on vintage lenses - I find that adding a bit more sharpening can help to even out the shots.

You'll have to have a word with the talent then!!

It is a blanket statement, but it's my experience. I've heard it frequently mentioned from others too. I shoot on the 200Mbps 10-bit ALL-I 1080p mode on my GH5 and it cuts really nicely, so the ALL-I definitely makes a difference, but it's not just the editing performance that I've heard pros say they appreciate, it's the aesthetic too. I am also glad it's starting to appear in more and more low-priced cameras.

I think the GH6 will have to be impressive to be a success. In this sense, it will have to really beat the FF offerings, which have caught up and overtaken the GH5. Unfortunately, people want things like you mentioned, which are better delivered by FF. The advantages of MFT are many, but you're just not interested in them. You want features that are better delivered by a larger sensor (47MP), are easier to deliver in a larger body (ND), are well catered for in the common sensors (PDAF), and you're willing to have something almost the size of an S1H, which is absolutely enormous compared to MFT. I wonder how much Sony have chosen not to develop innovative MFT sensors in order to drive people to FF where they can easily compete? It's a smart strategy.

I've heard some editors say that the trend of having more cuts is just lazy film-making. I can't remember where or I'd link it. The rationale was that cutting creates visual change, and makes things seem like they're moving and exciting. I think frequent cutting is used a lot in fight scenes. The example I saw was comparing a fight scene that had lots of cuts (can't remember where it was) with a fight scene from one of the original Bourne trilogy where Matt Damon fights a guy in Morocco (?) and it had a lot less cuts but was still really brutal. When you looked at the two scenes one after the other, the scene with more cuts actually had less action and less innovative camera angles and cinematography. Basically they were trying to pump up a weak scene by having you constantly trying to re-orient yourself and not notice that the content wasn't that exciting. If you're curious, there's a fascinating effect by having extremely long shots that seems different to normal cinema. I remember in particular that both Russian Ark and Roma had. Like it was a different kind of cinema, at least after it'd been a while since the previous cut. IIRC Roma had a few sections of relatively normal cutting with some hugely long shots.

I just assume that all content these days is steamed, so to clarify: I find triple-h26x compression (h26x in-camera, h26x export from NLE, and h26x streaming compression) to be harsh, compared with double-h26x compression (Prores in-camera, h26x export from NLE, and h26x streaming compression) The acquisition compression matters more than the export and streaming compression as often it's in a flat format, so any digital nasties get amplified with all the processing (adding contrast and saturation, sharpening, etc). Prores was designed as a professional digital intermediary, and the implementations in cameras will have been honed over many years to be as visually optimised as possible. The bitrates will also have been selected to be generous, as professional video environments aren't so worried about a 10 or 20% increase in file sizes if that's going to mean a difference in quality. The H263/4/5/6 standards were designed specifically for lower bitrates: Source: https://en.wikipedia.org/wiki/Advanced_Video_Coding The overall aesthetic quality of h26x family is secondary to bitrates, because unlike a post-production house, Netflix or Prime might get bankrupted by a 20% increase in their total storage, processing, streaming infrastructure and data costs.

Still at it, but now looking at TV episodes. In contrast to the short YT films, here's the analysis from a 40 minute TV episode: Almost 2000 cuts! The way this works is that you're analysing a clip BEFORE adding it to the media pool, and when you add all the clips into the media pool you can't change the start and end points. So during the analysis process you can further cut up these clips as they appear on the timeline, but you can't join some together. I'm not going to review these manually, and there will be false-positives (it thinks the arrival of a lens flare or other motion is a cut) and false-negatives (a jump cut that isn't too visually different) so there is no threshold that will get them all right. As such, I take a cautious approach and set the threshold high, so it misses real cuts but doesn't cut up shots on big movement. This way, if I'm analysing a part of the timeline I can further cut it up manually, but I won't be stuck with cuts that shouldn't be there. This approach creates about 1000 cuts, which is about a cut every 2.4 seconds. Not too far off really. Then I can start sorting and in the parts where I want to see every cut I can manually refine at that point.

Also, I'd imagine you'll probably really like FilmConvert. I've spent a lot of time learning colour grading and I value doing things manually to deliberately build my skills, but if all I was interested in was getting a good result I'd just buy one of those packages like FilmConvert or Dehancer etc and be done with it. I wouldn't recommend them to everyone, as film-emulations aren't the only look so they're kind-of a one-trick pony, but if you like the look then they're definitely the fastest way to get from SOOC to done.

If you're using a strong look like a film emulation then it's worth noting that the preferred approach for colourists is to apply the look and then grade the clips with the look applied.. "grading under the LUT". This will prevent you from having to make two primaries passes, and will save you from adjusting things that are visible without the look but aren't with it. If you're skeptical, grab a bunch of clips and time yourself doing both methods, you might be surprised.

I think the challenge in the MFT space is the limitations of the sensor. The GH5 and P4K basically took all the modes the sensor was capable of and enabled them, the GH5 compressed and the P4K in RAW. If the best MFT sensors are limited to (for example) full-readout 60p and FHD 240p, have dual native ISO, X stops of DR, and a rolling shutter then you can't make a camera that exceeds these specs, so the best you can do is give good codecs and features and make it reliable and in a good form-factor. To be really significant, the GH6 will have to do something groundbreaking, which would require an innovative sensor or crazy processing of some kind. Still, a camera that combined the best specs and features of the GH5 and P4K would be pretty good and probably get my money. Yeah, those codecs are particularly weak. Is Olympus known for improving the codecs in firmware upgrades after the fact, like Panasonic did with the GH5?

Indeed! It's a fair point too, because for our purposes (where we'd use the camera rather than just mounting it and walking away) we need a rig that's usable so it's the rig size that matters rather than the box itself. For that the Komodo is actually a fascinating form-factor and brilliant in a few ways, mostly because of the built-in screen which can be used with a MF-style viewfinder: So you just add a couple of batteries, media, a loupe, a lens, ND and then you're off! The BMMCC obviously didn't have a screen so that made a rig essential and depending on how you rigged it basically tripled the size of it.

Ah, well RAW is RAW, so in that case they're directly comparable. For the compressed version you'd really need to do a visual inspection because 540Mbps isn't an amazingly high-bitrate for 8K. The uncompressed data rate for 8K 10-bit 4:4:4 (after debayering) is about 25,000Mbps, so 540Mbps is only 2.2%, whereas for reference the Prores HQ data rate is about 9% of uncompressed. Prores is visually different to h264 and h265, which both seem more aggressive and digital looking than Prores to me, so the 540Mbps might look considerably worse than a still from a Prores HQ file. Still, the proof is in the pudding so you'd have to actually look. A good way to preview it is to put it 1:1 on a larger monitor and then gradually move away from it until you can't see the pixels anymore. Then you can work out how large the image would be when printed that size and how far back you had to move, then scale that to how close people would stand and you'll get a rough indication of the upper limit on size.

All else being equal that's true, but it'll depend on the compression, which apart from RAW and Prores is typically very high. A 100Mbps 8K frame won't be as good as a RAW 4K frame.

The weight for the Takumar wasn't from a completely reliable source (you don't really find specifications sheets for vintage lenses straight in google) and I wasn't sure of the version either. Takumar do strange things with their versioning - "Super Takumar" and "S-M-C" and "SMC" are all different IIRC, and who knows if it varied from year to year either. Lots of things lost to time around these points in history I think. I have plenty of way heavier things, so I'm with you in terms of things being heavy. If it was heaviest setup of only camera, lens, and adapter, I might give you a run for your money - the GH5 and something like the M42 SB and 200mm F4 Minolta would be right up there. Or others. I have several vintage lenses past 200mm I think 🙂

Pixel peeing here, but it looks quite compressed: Is that compression from the original file or the compression to upload the still? My understanding of wedding photography is that the occasional couple will drop real money to get a large print of the best image of the day to hang in their house - maybe 24"x36" or more. Does that still happen? If so, that's the worst-case they'd have to stand up to, probably being viewed from about the same distance away as its larger dimension. I do really like the idea of grabbing a still from a video, especially if you're a single shooter. I've played with printing out stills from video and my recollection was that bitrate mattered a lot more than resolution, at least for modest sized prints.