kye

As a 47mm F2.2 it's pretty close to a 50mm, so a pretty good FOV if that's the look you like. I've just bought the Panasonic 25mm F1.7, which is 50mm F3.4 on the GH7, and it's my first 50mm FOV lens, so I'm keen to start taking that out and about and "learning" that focal length. It probably seems odd to most that I've never shot with that focal length, but it's never really suited the situations I've been shooting in. Big and heavy isn't desirable, especially as I have some pretty killer combinations in that range already. I've got the Voigtlander 17.5mm and 42.5mm F0.95 lenses, which are 35mm and 85mm F1.9 equivalents, and with the Sirui 1.25x adapter they're as wide as 28mm and 68mm F1.5. Problem is those combinations are each 1300g / 4.6oz (2100g / 74oz with GH7) which is usable for hand-held shooting but my arms are pretty much done after a few hours. Of course, I'm also pretty much done with being on my feet for that long too so the setup isn't the limiting factor. These setups also have the advantages and coolness factor of being anamorphic too, so there's that! I'm slightly tempted by some other much stronger anamorphic adapters too, but in playing with the Sirui I've learned that they're very dependent on the taking lens, and some lenses work and other lenses that seem very similar are an absolute train wreck, so it's basically a blind purchase and these things aren't that cheap. My current preferred Night Cinema combo is the Takumar 50mm F1.4 and M42-MFT speed booster, and I bought the EF-MFT speed booster to try and get a slightly cleaner image, but unfortunately the 50/1.4 -> M42-EF adapter -> EF-MFT speed booster stack doesn't work as the back element of the Takumar interferes with the EF SB glass. I figured that the EF SB would open up a whole new world, as EF used to be the default standard for about half the worlds camera users, plus EF and PL were the de facto standards for cinema. My recollection was that there were tonnes of interesting and super-fast third party EF lenses, but now I go looking there doesn't seem to be so many to be found. If budget was no-option then there are a lot of cinema lenses that are very interesting, even anamorphic ones, but budget is a consideration so unfortunately things like the USD1400 Blazar Cato 85mm T2.8 2x anamorphic lens will have to just remain a dream! Apart from this being like me asking "can you recommend a lawnmower?" and you suggesting "just sell your house and move to a NYC apartment", I actually don't think it's true. I've done a lot of "what if" scenarios for what I'm trying to achieve, and while it might have worked for you specifically, when I look at the lenses for a given system I find every system to be very lacking. Sure, you can get fast lenses at 28/35/50/85, but there are all kinds of other gaps that MFT just doesn't have. This is gradually changing, as manufacturers gradually fill out the various lens lineups, but my impression of the current landscape is that most FF systems have the fast primes and holy trinity zooms (28/35/50/85 and 16-35/24-70/70-200) but very limited and patchy coverage of small and light lenses and in-between focal-lengths, etc. Adapting vintage SLR lenses gives access to character lenses, but only FF ones, not really S35 cinema ones (unless you go to crop mode which is throwing away resolution and now you're half-way to MFT!), and not the S16 ones. You sort of get the standard focal lengths in modern and in vintage, but that's it. MFT has all the small and light and in-between focal-lengths, but the weakness is in the fast primes and fast zooms. MFT used to have all the advantages of mirrorless essentially being able to adapt any SLR lens, but as time goes on, more and more of the interesting lenses are designed for mirrorless APSC/FF cameras and therefore not usable (or only available in MFT mount so you don't get a SB advantage) or are EF cinema lenses and are huge/heavy/expensive because they compromised on size and cost to ensure they're sharp sharp sharp sharp sharp enough for modern use. As an example, I have a tracking sheet of all the lenses I own (not every one available) and it's FF equivalent. This table includes: 15mm 18mm 26mm 28mm 30mm 31mm 34mm 35mm 40mm 50mm 53mm 56mm 59mm 64mm 70mm 71mm 75mm 78mm 80mm 82mm 83mm 85mm etc. I don't even think it's complete! Most of those are different characters too, being a mix of different manufacturers, vintages, and combinations of being native / with a SB / with an anamorphic adapter / with SB and anamorphic adapter, etc. So yeah, if you happen to want a fast standard focal length, or a fast trinity zoom (that's enormous), then FF is great, but the rest is pretty lacking. The other reason I'm not moving to FF is that most of what I shoot is on a native 10x zoom, which FF doesn't really have a good answer for. The other other reason is that the work for me to sell all my MFT equipment and re-buy in FF would make it so that I should just work those hours at my day job and increase my budget, so selling things isn't really cost effective. The other other other reason is that the GX85 has all sorts of configurations that FF can't match, so I wouldn't be selling that anyway.

Great post and lots of good points. Research on preferences is possible and there are actually some great resources out there talking about what aberrations are preferable, and are likely what ARRI have included in their colour processing. You have probably come upon these before, but for those who haven't, some of the resources I've come across are.. Color image reproduction of scenes with preferential color mapping and scene-dependent tone scaling https://patentimages.storage.googleapis.com/41/0a/e5/0a78ae57552549/EP1158779A2.pdf It's a patent obviously, so it's very dry and technical, but it contains interesting phrases such as: "a visual color reproduction of a scene having preferred color reproduction" "It is well known in the art, that the best reproductions of original scenes do not constitute a 1:1 mapping of scene colorimetry." "Pre-ferred color reproduction is defined as a reproduction in which the colors depart from equality of appearance to those of the original, either absolutely or relative to white, in order to give a more pleasing result to the viewer." There are other patents from earlier (1995 and 1996) from the same authors, so you can see them developing their knowledge. Visible skin color distribution plays a role in the perception of age, attractiveness, and health in female faces LINK to PDF It's a research paper, so technical again, but contains: "Here, we show that facial skin color distribution significantly influences the perception of age and attractiveness of female faces, independent of facial form and skin surface topography." "The results presented suggest that visible skin color distribution plays an important role in subjective evaluation of female facial beauty." These can seem very theoretical, but they have direct application in colour grading and development of looks etc. This video below shows how adjustments in hue compression and tone manipulation can make skin tones far more pleasing. FilmLight and the work of Daniele Siragusano are really the most advanced that I've seen (published publicly anyway) so are incredible references. Video (linked to timestamp): If you can't be bothered watching, essentially he takes the following image: then alters it to be very unflattering by adding green under the eyes and increasing magenta on her nose and cheeks: but then applying a film look, the rendering of colour almost completely removes these unflattering alterations: The film look reduces the magenta/green in that area of skin tones, which is why the rendering is flattering. This is the film rendering of the image, but without the unflattering alterations: Obviously the image is nicer without the alterations, even with the film look applied, but the alterations are far less unflattering with the film look applied rather than just with the normal 709 style grade. This is a very simple example (compression of the magenta/green axis in the skin-tone area of the colour cube) and there are others too, such as the compression of the hue outright, to reduce the variation between yellow and red in the skin tone area. For example the ARRI Alexa 65 promo video shows a range of people and skin tones, including this older gentleman: Obviously this whole image is desaturated, but look at the variation in tone - there's a lot less than in real life, but it's in a way that is flattering, and it seems completely natural in the context of the video, where the whole video is quite muted: but compare that to the amount of skin tone variation that people actually have when they age, the lips and areas around the nose and eyes etc are far more red and contain far more yellow-red variation than the above image shows: There are all sort of other things - that's just two. Going back to the Kodak patent and seeing what the "preferred colour reproduction" is doing, we see it's doing all kinds of crazy stuff: This is a plot in L*a*b space, so is essentially the colour wheel (luma is the axis not plotted). The symbols are where the transformation puts the colours and the line points to where the colour used to be. So it's taking the more saturated colours and making them more saturated, but not doing this with the less saturated colours, however you'll note there are some exceptions in there. This is happening in a 3D space though, and some adjustments will be happening dependent on the luma values involved. HOWEVER, and I just realised this, because the Alexa outputs don't do luma-specific manipulations to hues, any such luma specific adjustments must be in the ARRI709 LUT, not in the camera. There's tonnes of stuff in here, for example, the paper shows this diagram - Figure 14: and the description of what it shows is: "Fig. 14 shows how the hue reproduction could be modified for a system including variability, so that the optimum system color reproduction is obtained including all the sources of processing variability. In this case, the memory colors skin 60, sky 58 and foliage 56 are consistently and smoothly moved towards a hue line, while yellow hues 62 are shifted towards orange." It doesn't talk about skin specifically, but the a*b* space is this: So the skin is in this area of the plot: Which doesn't seem to involve any hue compressions or alterations at all. I'm not sure why they omitted those. Anyway, fun stuff, but a very deep rabbit hole indeed.

I bought an EF-MFT speed-booster, now I need some lenses! It's a VILTROX EF-M2II, and I tried my 50mm F1.8 and the AF was somewhere between functional and like it had a mind of its own. The focus ring on the lens has such a short travel that it's impossible to use as a MF lens (which it wasn't designed for so I understand). I'm looking for some interesting / fast / cheaper EF lenses that will be fine manually focusing. I'm especially interested in lenses F1.4 or faster but are still budget lenses. My thoughts immediately turn to the early days of cheap Chinese glass, but I can't seem to find any as any search is overrun with Canon AF lenses, or with new MF lenses that are for EF-M mount but are incorrectly labelled as EF (even Meike has a lens labelled as EF that is clearly a mirrorless lens!), and while EF was the only mount people never bothered to include that information and so things don't appear in searches. Laowa doesn't even let you browse their products by mount! Our new robot overlords suggested some options: The Samyang/Rokinon series: 35mm 50mm 85mm F1.4 Zhongyi Mitakon Speedmaster 50mm f/0.95 (nice but it's over AUD1000!) Zeiss ZE series: 35mm f/2, 50mm 85mm f/1.4 MEIKE 50mm F1.2 Anything else I should be looking at? I really thought there were more third-party options around than just those.

If you like.. the thread I linked to included a link to a comparison of dyes and the digital sensors all had significantly weaker dyes in the bayer filters, probably to increase the sensitivity to light, but it means the spectrums of RGB overlapped a lot more than film, which hardly overlapped at all. This has all sorts of implications on digital vs film, and I have been pondering it since linking to that, but I'm not really sure what it really means in the context of ARRI colour science or even good colour science etc, which is why I didn't mention it. One thing worth mentioning is that when it comes to additive colour (like light) the response is a cone that gets wider (more saturated) at higher luma values, whereas subtractive colour (like dyes) the response is widest at the lowest levels of luma. This gives you two cones where the additive one (which is any monitor) can't fully replicate the most saturated parts of the subtractive response. However, now we have much higher dynamic range devices (HDR displays) they pretty much cover the dark-and-saturated part of the subtractive cone, and considering that modern cameras also have this kind of dynamic range they also capture it too, so that problem is mostly solved, with the only wrinkle being that you have to compress the whole gamut when grading your footage for SDR displays so it all fits, which is why film emulation plugins are so useful because they do a great job of this.

The "Australia Tax" is definitely real. It's getting less now thanks to how easy it is to buy direct online from overseas, but if you buy locally you pay a significant premium to the local importer who has exclusive rights to set pricing and sell those goods. I've known people who needed to buy something significant for their work, so rather than buying it locally, they took the family on a weeks holiday to Singapore (or the US) and bought the thing there and brought it back in their luggage... and even including all the costs of the family holiday etc it was still cheaper! I also remember someone from tech company needing to download a huge database from overseas and with internet speeds at the time they ended up flying there, burning it to a bunch of discs, then flying home again, and if you divided the total amount of data over the entire duration of the trip it was still dozens of times faster, and might even have been cost competitive on the data rates too.

Apparently they can't even educate their own sales reps about their own products, so their ability to sell a camera that doesn't have the latest specs may be less than required. Emily said one thing in the video that was amusing, she said she had to hand it to Sony, as they have managed to brainwash their customers into thinking that larger sensors are better and that smaller formats aren't useful for anything. This is true, there is all kinds of misinformation out there that people are just constantly repeating to each other. The total amount of money spent on marketing must be absolutely unfathomable, but the thing is that companies will gladly pay for it because sadly, it works.

My understanding is that WB is Linear gain, so it makes perfect sense for it to be before debayering. The logic is sound - WB is a mathematical adjustment to the sensor read-out and needs to take place before any creative operations, which for ARRI includes the debayering operation itself. Photographers click the button and expect the file on the card to contain the RAW un-altered readout from the sensor, which they think is somehow pure and have all these rules about what you're meant to do that they all just parrot to each other without anyone actually testing them and actually learning anything. ARRI is light-years beyond this, doing all kinds of processing. As someone who learned photography and then learned video, my impression was that going from photography -> videography was harder and took far more work to understand than learning photography from scratch. I randomly found a thread many years ago where someone posted the five most expensive photos ever sold without revealing where they were from, and the reaction was complete derision, with people saying things like "out of focus" "not sharp" "wrong WB" etc. When the OP revealed these were sold for millions of dollars each the reaction was various combinations of "we are right and people don't know what a good photograph is".

Emily talks about the L10 and MFT small cameras: The main point I took from this made complete sense to me. New cameras need to have new specs, new specs means current sensors and they need more power (both for the sensor and also processing much more data), and therefore there is more heat and larger batteries are required, therefore the chassis just needs to be larger. I note that camerasize.com now has the L10 in their database, so that means we can get comparisons! It is larger than the LX100, but not by a crazy amount... Emily did note that it's not pocketable, but also that after carrying it around Japan for 10 days she didn't feel that it was heavy or bulky at any point, so I guess there's a distinction there between pocketable and it feeling large / cumbersome / etc. It's also interesting to compare to the S9: Obviously the S9 needs a lens (unless you're an abstract filmmaker!) but the purpose of the comparison is that the S9 is a FF MILC camera with IBIS and the body is slightly narrower(!) than the L10, and the same height. So, I see no reason there couldn't be a GX100 in the S9 body as it would be the same except with a smaller sensor. Obviously the depth is different, but that is lens dependent, and anyway, if it's not pocketable then it's down to how large it is when it's in use, and at this point it seems the L10 grows more than Pinocchio running for President: I'm sort-of hoping for a GX100 as it would be extremely tempting, and also hoping it won't happen, as it would save me from being extremely tempted, and draw my attention away from my GX85 that (apart from it gradually turning into a film camera) I am remembering how much fun it is to use!

Yeah, there are differences for sure. Some situations show larger differences than others. One theme is that the Panny versions are much more saturated in skin tones than the Alexa. Coming from the kings of colour science, this has got to be deliberate, so even ARRI aren't trying to match the Alexa. I think that instead of asking "are they identical", the better question would be "does this impart some magic". I mean, everyone agrees that ARRI has the magic but their cameras don't even match each other between sensors, so it makes no sense to have a higher bar for the Panny version than we'd apply for ARRI themselves. My impression was that the ARRILogC profile and the ARRI709 LUT is the most like the Alexa, with the V-Log -> CST -> ARRI709 LUT not being as good. This makes sense as whatever colour / gamma secrets ARRI applies in-camera can be put into that profile as it's customised to the Panny sensors and also safely locked inside the camera away from prying eyes (unlike anything they put in their LUT). However, the Alexa also responds differently in the spatial dimension, with the far-red response including a more spatially distributed response from skin tones. We also know that Alexas process the image spatially in-camera due to their texture options and processing. Who knows what is going on with texture in there. IMHO the texture of Alexa images is right up there in importance as their colour response. None of this includes temporal aspects either. While I struggle to think of what processing might be occurring in-camera between frames, there might be some (we can't tell), and that's beyond the possibility that the hardware itself has some sort of secret properties that contribute to the image. FDTimes did an entire episode on the Alexa 35, with interviews of over a dozen people and 100+ pages: https://www.fdtimes.com/pdfs/free/115FDTimes-June2022-2.04-150.pdf Here is the image pipeline in the Alexa 35 (page 59): To give some idea about how stunningly out of our depth basically everyone on the internet is who talks about this stuff, starting on page 116, Dr. Tamara Seybold talks about Textures.. "For example, the debayering already needed to obtain the full color image doesn’t only generate RGB values but also influences the perceived sharpness and grain rendering. And many more steps influence the clarity and grain that are important aspects of the texture of an image. So we, in the image science team, pushed hard to obtain the best results by really optimizing each and every step in the image processing pipeline, not only for the best color rendition but also for the best texture, as we call it. We did that in a holistic way, optimizing steps in the beginning of the pipeline together with later steps so that the overall result would be best. At some point, this came down to having more than 30 parameters that we had to optimize together—a huge amount. We specifically had to build a small “texture grading machine” to be able to optimize all these parameters together." (emphasis added) I don't know about anyone else here, but I would struggle to even list 30 parameters, let alone identify all the parameters, isolate the 30 that matter, then find the sweet-spot (or sweet spots) in a 30-dimensional space. This is regarding the Alexa 35, but I remember reading in there somewhere that the innovation of the Textures feature is that you can choose different profiles on the new camera, whereas on the old ones you only had the one, and that on the previous models they had chosen a texture configuration that was their best attempt at a one-size-fits-all. So the inference was that the previous cameras were also doing this kind of processing. By implementing their colour science inside the camera, they could be doing all sorts of stuff. They could have things that analyse the image and then apply different treatments depending on the scene the camera was capturing. They certainly have a team capable enough and a camera with enough processing power to have a dozen, or a hundred, or a thousand, LUTs or algorithms inside it and be changing these things based on context or WB setting or sensor temperature or whatever the hell else they found was useful. The sheer depth of knowledge that has gone into their image science is incredible. In 2009, Glenn Kennel joined ARRI as their CTO, which was a new position at that time, and in 2010 he was promoted to President and CEO. Glenn had previously worked for Kodak from 1980, and worked on various things that involved the gradual digitisation of the pipeline, including things like telecines and film scanners etc. My understanding is that his contributions at ARRI were pivotal for the development of the Alexa, which was the first digital camera to gain wide acceptance within the industry and did so due its film-like response. A bit of searching revealed some interesting discussions we already had during lockdowns..

I'd imagine it'll be somewhere between exorbitant, ridiculous, and "this is just an action camera - WTF".

Yeah, nice images! What I have learned in all my research is that there are tiny little sweet-spots in every aspect of the image. People lust after OLPF filters with just the right amount of softening, people marvel at skin tones with just the right amount of compression and hue manipulation, or lenses with just the right amount of distortion. Each of these imperfections / distortions / non-linearities has a certain feeling and aesthetic and comes with a range of associations. Combinations of these will have synergies, or won't, or will clash by pulling in different directions. When we moved from cameras that shot in the publishing resolution and recorded colour in the publishing colour space and gamma to cameras that had higher resolutions and log colour spaces etc, we went from looking at the image that professional imaging scientists tuned into a final commercial product to taking an image that was designed to be manipulated and then applying our own (likely far less skilful) texture, colour, contrast, etc. It's taken me a good decade to start reliably getting images I like, and even then, I'm using a film emulation plugin that is doing most of the heavy lifting.

100%. ......and if you switch the question from "is it visible" to "is it important for the content of the video" the answer gets even clearer!

Just thinking more about this, in a studio setup where everything is controlled the 'weaknesses' of older cameras often cease to be important or even relevant. IBIS doesn't matter, DR is irrelevant as you can just adjust lighting, size and weight don't matter, AF doesn't matter (and isn't desirable as the last thing you want is it focusing on your hands and the background whenever you move around), etc etc.

To add to the above, Matti Haapoja (perhaps the king of pixel peeping "cinematic" YT) uploaded a bunch of videos that were edited in 1080p (and upscaled to 4K for export) and he tracked all the comments and not a single person commented that the videos looked different or whatever. In a blind real-world test literally not one of the pixel-peeping techno-fetishist pedants could tell. Of course you'd still want a good 1080p image, some 1080p cameras were better than others. The added bonus of this approach is you only need 25% of the computing power to edit it. Or if you go with h264 instead of h265 then it's even less!

How interesting. I guess something like that would be very difficult to determine for cinematographers etc, but if you're a lens manufacturer with all the test equipment then buying all your competitors lenses and then testing them and taking them apart would be pretty standard operating procedure.

There's a touch of gate weave in there, but I can definitely increase it. I was surprised about the grain actually. The amount I had dialled in was about 0.16 and I uploaded a test sequence that had 0.1 / 0.2 / 0.3 / 0.4 and the 0.2 was clearly far too much, so the above was just 0.1 as I figured if it's too little then it would be similar to when people upload and YT compresses it a lot and removes most of the grain. I guess there's two approaches.. The first is to make it look like film in the NLE and then upload it and YT will remove lots of grain but it will look like real film uploaded and processed by YT. The second is to do whatever you have to do in the NLE to make the final YT stream look like film. The more I look at my references the more I realise there is incredibly variation between them and that I have tonnes of leeway.

OK, now I get it. There's this one, the PRO, and then the ILS which has interchangeable lenses. I'm waiting to see what that one is like. I love me some lenses!

Round 4. Changes: Included lens emulation Stabilised shots Lowered grain I completely rebuilt the grain nodes in a different OFX trying to refine it, then realised it didn't animate. FML. The lens emulation includes adding a vignette, softening the edges, and adding a slight barrel distortion. The grain seems a lot on certain things (like the sky) but doesn't appear at all on other things. I see no pattern for it, but this is how real S16 films also appear on YT so I'll leave it to the judgement of those with a better eye than me. These are my settings for Film Grain OFX - it appears there's quite some adjustments, so let me know if I should play with anything...

Very interesting. I've been looking into this for a long time and hadn't heard some of these insights. Very useful. One thing he got wrong, at least for v3 of their colour science is that there isn't anything luma-specific done to the image inside the camera, it's all done in the LUT. The evidence for this is when people do under/over tests, where the camera is deliberately under and over exposed, when you correct the image in post to the correct exposure the colours are all the same. If they were warming the highlights or cooling the shadows in-camera then when you overexposed and brought the exposure down in post then that warmth would be baked-in and your mid-tones would be warm (or they'd be cooler if you underexposed) but that's not what we see. It's also no secret that they compress the skin-tone hue range and also tend to skew yellow, especially compared to Canon which skews magenta/red. The IR-cut filter letting in a bit of far-red so the skin-tones get that scattering is interesting. To a certain extent it might be 'recoverable' in post (ie, perhaps we can guess what might have been there based on what info we do have). Perhaps the key aspect of any such attempts would be to blur this new channel once it's been simulated, as this is the information coming from deeper in the skin and is scattered a bit. It's been hinted at that part of the 'Cooke Look' was that they used materials that slightly blurred light at a range of frequencies within skin tones, so the lenses sort-of worked like a skin-hue-only diffusion filter. Potentially anyone with a full-spectrum camera (OG BMPCC BMMC anyone?) could seek out IR-cut filters that are designed to let in more far-red. Potentially even people with IR-cut filters on their sensors could get strong ND filters to boost the relative proportion of IR coming into the camera. I'm guessing that the right amount of the right ND might do it - have enough ND that the IR is boosted but not so much that the blacks become polluted. It's funny he mentioned that the iPhone has a strong IR-cut filter - I was testing my iPhone 17 Pro the other day with lots of ND (because it doesn't have an iris!) and I got a ton of IR pollution, including getting a non-trivial amount of it when using the same vND I use on my GH7 which I've never seen any IR pollution on. So the iPhone must have less IR filtration than the GH7. It's worth adding that a lot of these things are also accomplished by film emulation.

For the talking-head stuff, almost any camera will be good enough if given enough light, so I'd suggest you concentrate on getting 1) enough light so your camera is at its native ISO, and 2) lighting that is flattering and creates depth and contrast in the image. There are lots of videos on YT that show this, and the before/afters show what is possible. You don't need expensive lights either, there is tonnes of info on home DIY hacks using lamps and cheap shower screens as diffusers, etc. The standard approach is 3 Point Lighting, like this: This video is a good primer and talks about how to use (or avoid) existing light sources like natural light and ceiling lights etc. Other videos that might be useful: This video is longer but starts with a complete setup, so acoustics etc too. Cameras get all the attention, but in the real world are some of the least important parts of the whole setup. You're lucky in that you're building something indoors for one specific use in an environment you control and (hopefully) doesn't have to be portable and easy/quick to setup and pack away. With a bit of effort you should be able to get a great setup that works really well and doesn't cost much at all.

Thanks for the feedback! I'm also happier with the images, but I think there is still more juice we can squeeze. My attitude is to assume the aesthetic impressions people have are correct and to try and reconcile those with the technical process, especially when the two are at odds with each other. The new texture approach to emulate a neg stock and then a print stock: Grain added at timeline resolution (full sharpness) with Resolve Film Grain OFX Gaussian Blur added to image+grain so that the sharpness of the image and grain match Grain (same as above) Blur (same as above) I think this more accurately mimics the architecture as with film the grain IS the image, so applying a softening to the grain or incoming image before they are combined would risk creating a mismatch. Then it's done a second time, which makes sense as if you print an image from a film stock onto another one, the image coming out of the print stock will be the print stocks representation of the neg stocks representation of the scene. The Resolve Film Grain plugin is a bit odd, with expanded/different controls than the ones in the FLC, but it lacks the presets I wanted so I had to roll my own. I think the grain was pixel-sharp, but there's a chance some softness is coming from there. Also not sure if Gaussian is the right flavour of blur. Let me know if there are better options.. Thinking about your impression that the image was sharper than the grain brings me to think about what might be making the image too sharp. The first potential culprit might be internal sharpening in the GX85, but I don't think this is the case. I expected that it would have sharpening, and some time ago I sat down to test how much it had and then design a blur that would counter it, but when I compared RAW and JPG stills (4592px wide) vs the 4K h264 on a 1080p timeline, I found no discernible difference, so concluded that the downscale from 4K to 1080p was enough to eliminate the (inevitable) sharpening that will be in the h264 stream if viewed at full resolution. The second potential culprit is that this image pipeline still lacks all the softness of anything in-front of the film-stock, specifically lenses and filtration. The 14mm F2.5 on the GX85 is pretty sharp, even when shot wide open, so might not be typical to the look of S16 from back in the day where the images might not have been so sharp, or they might have been sharp in the middle but not so much towards the corners? I have done a bunch of lens emulation stuff before so could easily reach for some of that, and I should add it anyway, just because it's more in-line with the overall goal of the project which is to evoke a more gritty retro mood, rather than to realistically emulate S16 film as if it was shot on modern lenses. Their vintage ND filters probably weren't quite optically perfect either. I should also do some tests to work out YT compression so I can start sharing moving images, as grain etc is completely different when moving to when not. Is there anything else I should be looking at for Round 4?

The GoPro one in Seoul is very impressive.... I'm guessing they must be doing super-duper processing of the image. Still, for the form factor it looks like a great result. I'd be curious to see how the low-light is compared with previous normal GoPros. Some time ago I made the case that an action camera is the perfect vlogging camera, as it's the size of a pair of wireless earbuds but is mostly without the issues that continue to plague the "vlogging" cameras that aren't wide enough / crop for stabilisation / etc. If I was a vlogger I'd be using one of these 8K action cameras for sure, using the wide for talking head stuff and cropping in for more normal FOVs. No-one on YT can tell if you edit a 1080p and upscale to 4K on export, so a 2K crop from an 8K sensor gives a pretty useful FOV, assuming the readout isn't terrible.

How impactful / costly would it be to get a camera without IBIS? If it's just a matter of buying another body to add to your existing ecosystem then maybe it's worth it?

Equipment with stabilisation like cameras with IBIS or lenses with OIS can have very different reactions to motion. For example the normal IBIS mode on GH5 vs the IBIS 'tripod mode' where it tries to eliminate all motion. I've been aware of this for a long time as one of the OIS lenses on my Canon 700D refused to let me pan the camera, and would hold the composition steady and when it ran out of travel it would jerk suddenly. It felt like the frame was 'stuck' and I had to 'pull hard' on it to get it to unstick and then it would jerk into a new position (that it would then wrestle with me to try and hold that new position). The jitters you're seeing from IBIS mechanisms may very well be some kind of undesired response to the motion of the camera perhaps? I have IBIS and non-IBIS cameras and I previously frankensteined a rig with four of them and then walked around the backyard with it. I can probably dig up the details / footage if that would be of interest, although the cameras are quite different (IIRC the non-IBIS camera was the OG BMPCC). I've sometimes wondered what an IBIS mechanism does when it's switched off. Does the sensor get held in place by the motors, or does it get disconnected and flap around? If I turn my GX85 upside down to get the battery / card out of the bottom I can feel something moving around inside it, and I just always assumed it was the sensor. I just grabbed it and did a quick test: - camera off = rattling inside when I turn it upside down - camera on / IBIS on = no rattling - camera on / IBIS off = no rattling I guess it's held in place somehow? In terms of seeing subtle jitters in the footage, does a touch of stabilisation in post fix it? If so, that might be the quickest / cheapest option. If you're doing hand-held work then you might really miss the IBIS. When I was using the OG BM cams for street shooting (BMPCC / BMMCC) the OIS lens did a great job of pan and tilt but OIS doesn't stabilise roll at all, so the image had very unnatural motion and I had to stabilise the roll in post.

Round 3. Changes include: More softening I'm currently aiming for a neg + print look, so that will be far softer than just a neg scan only. This is obviously adjustable to taste depending on what look I want. Grain is more saturated and with more blue and less red Grain engine completely rebuilt from scratch based on a grain-blur-grain-blur architecture More halation and bloom I think my previous versions were based on a 35mm preset where I adjusted the grain to 16mm but the halation and bloom were still at 35mm amounts - oops! I've applied it to some old images and some new ones. I have built a little set of S16 reference images from YT and also from PNGs people have sent me of S16 film scans, and in comparing these new grades I feel like these are closer but are in that strange space where they are too sharp and also too soft, where there's too much grain but not enough, etc. I suspect that either means these things are still missing the mark, or that there is so much variation in my references that I'm somewhere in the middle and therefore not close to the examples at the extremes. Feedback welcome...