kye

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...

Wonderful! Great eye and great images. Definitely in the direction of Koyaanisqatsi etc, not only with the images themselves but the shifts in theme too. Everyone should do themselves a favour and watch it!

Great stuff! I also find that when using 'lesser' equipment my brain can more easily switch over from "will this make technically great images?" to concentrating on the compositions and creative aspects. I suspect that our subconscious knows that footage from the latest cameras gets very detailed technical scrutiny but once the tech is no longer the current model the technological fetishism moves on and it's only the creative people left. I also find a strange satisfaction from getting great results from older equipment, and I have no idea why, so I just go with it. Other potential advantages of older equipment: it's potentially cheap to replace, so you can be less precious with it, even risking things like taking it in the rain etc any scuffs or scratches or wear marks can improve it's appearance rather than detract from it (people say "look at that old camera and how worn and beat up it is, awesome!" and also "oh no, I scratched my nice new camera") older bodies are often smaller older equipment is more likely to be metal rather than plastic - I dropped my GF3 once and it just got a little dent with no other damage! the lower DR forces you to expose better in-camera rather than choosing the exposure (and therefore the subject) of a shot, forcing you to be more decisive when shooting etc etc. Plus... if you own them already.. they're free!

Old cameras have a number of challenges, including: - weak codecs, often 8-bit low bitrate files - terrible low-light - dated colour science and no log profile (rec709 profiles only) - poor DR - lack of IBIS or EIS - etc At the time these were pretty significant challenges. Now they aren't the challenges they used to be, because fast lenses and film emulation assist with all these limitations. Let's take these one at a time. Weak codecs Weak codecs, including 8-bit low-bitrate files can be soft, and can be overwhelmed by motion. By shooting with faster lenses you render more of the frame out-of-focus and therefore the limited bit-rate only has to focus on a smaller percentage of the frame. Thanks to cheap Chinese optics companies, we are now awash in F1.4, F1.2, and even F0.95 primes. The soft image is now no longer a liability, because compared to our modern 4K sensibilities, even 35mm film is noticeably soft by comparison. This means that by adding film emulation you'll be softening those edges and smoothing over any subtle compression artefacts. Film often has a more compressed colour palette, pushing hues closer together in many instances, lessening the visibility of artefacts. It doesn't work magic, but every bit helps. Terrible low light Cheap F1.4, F1.2 or even F0.95 primes sure make a big difference after the sun goes down. That "fast" F2.8 vintage lens you were shooting on back then is 3 stops slower than these things now. That can really bring a lot of situations back from being unusable to being at, or close to, native ISO. Dated colour science and no log profile Rec709 colour profiles are basically a creative filter the camera has applied, and they often weren't that good. Film emulation takes that image and applies an incredibly large transformation over it, which goes a long way to hiding any imperfections the colour profile might have had. It's like if you put on a pair of rose-tinted-glasses, you can still see that things have different colours, but any subtle differences aren't visible because the image has had a strong look put over the top. Also, film emulation plugins often come with controls for exposure and WB etc, which can help to grade the 709 footage, which was a major pain back before we had colour management pipelines. Poor Dynamic Range You know what else has pretty poor DR? Print film! Kodak 2383 has about 5-6 stops in the linear region, and then everything else in the image is squished into the highlight or shadow rolloffs. Yes, you can see into those rolloffs a bit, but if your camera has 8 stops then you've got at least a stop to put into each rolloff. People think film has huge DR, and it did at the time compared to consumer digital cameras, but it was the negative film that had the huge DR, not the print film. It's very common now for people to shoot on film, scan it, and then do everything else digitally, so they keep the full DR of the negative, rather than taking half of it and pushing it into the rolloffs. This is a still from Minority Report from 2002: It's not exactly a dynamic range demo - the streams of light INSIDE THE ROOM are blown out and every item of clothing the main character is wearing is crushed blacks. Lack of IBIS or EIS So there's a little shake in the files... well, film had this thing called Gate Weave, which was where each frame didn't perfectly align in the camera and so when played back there was movement of the whole image. Once we started doing digital intermediates people started stabilising the images digitally and that went away. When I went to the cinema and saw Goodfellas projected on celluloid they played a bunch of old ads and movie previews also on celluloid, and some were jumping around all over the place and some were rock solid (which means the projector the theatre was using wasn't the source of it) and much to my surprise, Goodfellas itself had quite a bit of it. By just using modern tools you can now stabilise things pretty easily, but this will create artefacts if you do it too strongly (especially if the camera had bad RS), but applying film emulation gives you much more leeway. This is because you can stabilise the image, then apply some Gate Weave, and once the viewer notices your images look like film they'll potentially just accept the shake in the image as being part of the film look. By adding Gate Weave and getting some grace from the viewer you can potentially increase the strength of the stabilisation you're applying too, with there being more wiggle room, and also because the softening of the image will mean that any distortions in the image will be slightly less visible. I was inspired to write this partly from my GX85 Super-16 camera project, but also partly by this video of the GH2 shooting at night. You can still see the ISO noise and macro-blocking creep in as blue hour ends, but he was also using the 9mm F1.7 and 35-100mm F2.8, the F1.7 is reasonably bright, but the F2.8 is pretty slow compared to things like the TTartisan 50mm F1.2 or the 7Artisans 35mm F1.2 primes that are $109 and $97 on B&H. These won't offer OIS, so your options for these on non-IBIS cameras are to spend more (Canon and Sony both offer 35mm and 50mm F1.8 primes with OIS) or to use a tripod or larger rig of some kind. Far from perfect, but much more useable than you'd think. These cameras have actually gotten better over time as the rest of the ecosystem is better able to support them. The only reason we don't think so is that our expectations have inflated faster than their potential.

100% Of course, completely irrational biases still exist. One of the major ones is that people seem to only want lenses so sharp they need to wear gloves to even pick them up, or lenses so distorted that it's like most peoples memories of Woodstock. There is a secret third option. These lenses have a subtle hint of the vintage vibe, but aren't completely out of control divas that need to be babied all the time. They're also cheap, freely available, have modern features like AF and OIS, and are fast and easy to use. To discover this secret, here's the thread where I talk about them, and reveal what they are and how to get them.

No worries! In terms of the lens, not a clue, but it really shouldn't matter.