Speaking of organic noise from digital cameras, I was blown away when I learned that The Holdovers was shot digitally. Apparently, they shot the Alexa at 1280 ISO to add natural texture in camera, before hitting it with heavy film emulation in post (with added grain/halation/gate weave). I was pretty much convinced that it was shot on specially made film stocks from the 70s, but it wasn't!
I did the same thing with a short I shot on the Mini LF. I pushed it to 1600. It’s Black & White, super high contrast. The noise pattern looks great. 3200 on the Mini LF looks like 16mm
I shot a video about this exact same topic that I just haven't got around to finishing. What you're talking about here is SO crucial for run and gun type filmmakers and I'm not sure why it's not talked about more on RU-vid. This adds to a handful of videos I've seen discussing it. RU-vidrs always say shoot everything at the camera's native ISO for the 'most dynamic range' and that is just so wrong. I would have saved myself a ton of unuseable footage if I had just known this earlier. I shot a short documentary last year and I got some of my best looking footage ever. Deep shadows with Zero noise. Completely blew my mind
I was expecting the usual 'the right iso' video, but instead you got a new subscriber! Very well put I would say! I would also add that there's also people shooting things outside of well lit sets, I find myself very often shooting wildlife, so very long zooms, high F stops, on a Micro 4/3 sensor because of the reach and the convenience, most likely while being in a dark forest 😅. I have to use different shutter speeds and iso than the 'correct' ones because otherwise it would be better if I stayed at home 😂
I agree with the concept of adding noise but you explained the exposure physics right. I'm a computer engineer, it's more technical. The ISO is how you set the range of the data captured by the bit depth of the camera, for example 8-bit, 10-bit, HLG, SLog3, etc. When you are adjusting the gain, you are setting the minimum and maximum range of the light you're capturing, which you can refer to as clipping the light on the sensor and normalizing (i.e. setting from 0-100%) to the bit depth. You have have heard the terms Expose to the Left/for the shadows, and Expose to the Right/for the highlights, and expose for the center/shadow and highlights. The light that is coming in is in a certain number range, and you need to scale that range so that you capture the dynamic range that you need. You do this with a histogram. You will see in the histogram that the light coming in at the different frequency ranges, and you need to get all of the information into the histogram that you want to capture. You can adjust your exposure compensation, and this will shift the light pixel data to the left or right of the histogram, hence the name expose to the left or right. If you are using 8-bit SLog3 in sunlight, you will either have to expose to the left or right with SLog3, or turn of log and expose to the center, but you will not be able to recover shadows or highlights. You are using the gain to set the range of the 8-bit and 10-bit color data, which means each pixel is either 24-bits (8-bits times 3) or 30-bits (10 bits times 3), but they pack that up in YUV format to get 4:2:2 or 4:2:0 subsampling, which is confusing, to do the Hybrid Log Gamma and SLog2/SLog3/CLog/FLog/etc All of them are variations on HLG, which is part of the HDMI standard but none of the other brands log curves are HDMI standards.
Great video Mitchell! More and more I find I'm having to rework my mind when it comes to how I should observe/handle ISO. This is not only a great reminder but an addition to how I think I'll view ISO moving forward. Grain can be good!
All this information is correct. Nice work. ISO is not about how good or bad a camera is … it’s about how good or bad a cinematographer is. The original ARRI Alexa at HD is still the best image for that film look.
best iso video ive ever seen! You got a follower more! Now i finallyy understand iso... photography (and hopefully cinematocgraphy in the future) will step up now. Thanks man!
I really enjoyed this video. As a Videgrapher I’m trying to level myself up and just purchased the FX3. I’m really obsessed with light and how that plays an integral role in my story telling. I did this after my RU-vid video colours and lighting were just waaaay off. I’ve been practicing and this knowledge you shared helped someone like me more than you know. It might be common knowledge but that you. I had no idea.
Another great video, anytime I see someone sharing the gospel on ISO I get feels. Only one note I would add... ISO does NOT change the sensitivity of a digital sensor, native or dual native sensor are fixed.... In general these "native" ISO's are where the camera manufacturer has best tuned for dynamic range and noise patterns... they are not actually more or less sensitive to light... Hence Gain... It's still the same process, ISO just became adopted after the DSLR revolution. Also The ISO is going to change from manufacturer to manufacture... DO NOT TRUST ISO to be accurate to film equivalents. Other then those points I totally agree, RU-vid has had a massive impact on the way DP's (especially in the last decade and a half)... the obsession with clinical images seems largely of a byproduct of test footage/review culture... Film was never clean and honestly go back and look at something like Heroes/Lie To Me from the late 2000’s would be considered by many "unuseable"... an at the same time the photo chemical process had reached the ability to deliver film stock and printing techniques that would be considered sterile by many now days chasing the "film look". I think most most great DP's will tell you its about your choices, your process and if you are working with a DP, its about what the two of you are looking for... not about "The right way"... This also goes for the current ongoing obsession with "perfect soft light".... but that's another topic.
@@Vitaphone don’t even get me started on soft light, mostly because I heavily disagree with the over reliance on big soft light. it’s good to know how to achieve it, but it’s robbing you of the creative choices a DP has to make if you use it exclusively in every scenario. Either way, thanks for sharing. That’s important info. Anyone in these comments please reference above in addition to the video!
@@mitchellbrinkerdp would love to start seeing more working DP’s on the RU-vid space opening up on the soft light/ “perfect exposure” topic… it’s been wild to see commercial and commercial photography lighting ideas dominate a lot of the conversation in the post RU-vid era… Most DP’s I work with with years under their belt skoff at this, but newer generations seem obsessed with the RU-vid meta.
When I color for people, I often get questions of whether they should target a certain ISO and over/under exposure level. I've simplified it to "your camera has one ISO its always recording at no matter the setting (or two if its dual native) and one white balance. Everything other ISO is a level on the volume nob. If it's not well exposed at native, it's not well exposed at all."
@@Lathrop_Media what do you mean? Outside of native ADC reading, everything else is gain functions. Even internally, the camera will just perform analog or digital gains on the sensor info when shifting ISOs or white balance. Peak pixel well saturation is fixed and sensor read noise level is fixed, so your floor and ceiling is static, assuming the ADC bit depth doesn’t cut off information. Outside of noise reduction (a lot of Sony cameras do pre and post ADC noise reduction), nothing changes that floor or ceiling.
Noise used to be a huge problem back when cameras were mostly using compressed codecs. But now with less compression and even raw internally in some hybrid cameras now I don't think it's as much of an issue.
Wow I had no idea about the more details in the shadows and highlights part of ISO. Thank YOU! Now I’m even less afraid of making a video in my room at 3200. Some people do that on purpose! 😂
Some good stuff here. Important to note though that how ISO changes above/below middleG distribution changes depending on the camera/brand, etc. you're using. Clog2 on C200 for instance caps the distribution at ISO800 and allows you to go below 800 and distribute down into the shadows but going above 800 doesn't actually shift into the highlights but it remains flat all the way up, only increasing the gain. So you only have 800 and below on that camera to redistribute if you want vs. other cams which tend to move the distribution above/below middleG up or down. Anyways, always good to test this stuff and understand how one cam responds under what circumstances. Noise doesn't bug me at all really, compression making noise look worse is sometimes the culprit or why I think "yuck." A little chroma noise reduction can go a long way too vs. trying to scrub the noise out.
@@mitchellbrinkerdp Yea it really depends on the cam. The gain situation on C200 was also why most people found themselves frustrated by the noise at the supposed base 800, because it already had 12db of gain, while ISO200/400 was 0db and 6db. Google "clog2 iso eoshd" in Google Images and you should find the chart.
With Sony cameras in EI mode ,EI ( exposure index ) its not ISO , your recording will always be in one of the two base ISO,s. EI is just like rating a film stock different than whats on the can , that is setting a light meter to 250 ISO on 500ISO rate film stock . Sony EI works like this too. If your base ISO is set to 800 , and you set the EI to 400, viewing with a MLUT , your image will get 1 stop darker from the correct exposure , so add light or open up 1 stop, you are intentionally setting a uniformed 1 stop over exposure off set , because your recorded image will still be 800ISO, and will look stop over exposed in Slog play back. So it's a bit tricky and counter intuitive , raise the EI , and you will get way under exposed recorded footage . Viewing without a LUT you will see no change in the image at all ,changing the EI.
Nice video man😃I have the fx30 with dual iso and “should” I still use the higher base ISO for outdoor and vice versa or does it not matter if they have 6 stops each direction either way? Also would’ve been cool to see the last clip you were talking about🙂
Only true uncompressed raw behive like this. Worked with many cameras. Only Alexa cameras, sigma fp and kinefinity cameras and cameras that shot true raw behive like this. They all shot 12 or 14bit raw in 4:4:4 color subsampling. On my sigma fp if I switch to 8bit h264 rules doesn't apply. Even in 10bit 4:2:2 gets tricky 😊
Isn't the optimal approach to use one of your camera base ISO's (though the ridiculous 12,800 on the FX3 is actually noisier just because of quantum fluctuations- physics!) and use ND filters and/or lighting (or light control) to make that base ISO 'correct' ?
@@muuuuusic6269 definitely does, especially since the A7S3 and FX3 have the same sensor. If you go to the 12800 base iso, you won’t have anymore information in the highlights. You’ll have the same 6/8 in the highlights and shadows as you do at 800. If you raise the iso from either base, you will get those stops in the highlights.
does this apply to photography with actual cameras ? if yes, my understanding is that rising iso and adding nd filter I could get better highlight roll-off ?
I just asked that myself after watching this video, so I did a quick test in my garden. Tested at ISO 64 & 2500 on a Nikon D850: no difference in highlight-recovery. Both are equally good. Only drawback: recovering shadows at ISO 2500 compared to 64 makes your camera look like from 2005 (or some 1600-speed film haha). Just as expected. I'll just stick to choose ISO-settings how it always used to be.
@@Juhkwez I googled the individual ISO/dynamic range charts for the camera. It’s hit or miss whether there is one. Or maybe I couldn’t find them. For the FX30, I think you’d be safe using the FX3 chart for reference. That one’s a bit easier to find.
