Now that options are available, thoughts are turning to quality and needs.
In stills you have RAW. All cameras shoot in it, better cameras allow you to access it and life is good.
From RAW you get jpeg or TIFF files.
Easy.
In video, thanks mainly to RED cameras, you do not get true RAW, because they own enough of it to make using the term and tech prohibitively costly as several companies trying to sue them have found out. Video also has to balance quality to storage realities.
Video therefore has “layers” of quality, best explained by Gerald Undone,
https://www.youtube.com/watch?v=wX9KGRHaMEY&t=42s .
Apart from the now dated camera specs being talked about, all this above still holds true, the main difference being direct to SSD recording from camera.
Internal limited users, tend to use LOG as their maximum quality level and those who break the shackles go into ProRes, RAW or similar. My own journey was one of liking standard profiles, feeling a draw to LOG, but compromising at Flat profile.
Ok, that is the meat of it handled, now the garnish.
First lets look at file sizes, storage and realities.
This is the key I feel. Having the best quality available to me (B-RAW), several levels down previously unavailable (ProRes and All-i) and even 6k Long-GOP, realistic storage is the first issue. Do I really want or need to come back from every job with 50-100gb or even more eaten up? I shot 5Tb of stills last year, but dumped 80% of that as unnecessary. My video was limited to occasional 1080/10 bit/422/Long-GOP, but even then a 2Tb drive was swamped in no time.
Realistically 1GB/minute is plenty, maybe more, but not very often. 1GB a minute is Long-GOP/10 bit/422/1080, so not much room to move. Ironically jumping all the way up to 1080 5:1 B-RAW does not add much more strain, but some of the intermediate steps do eat up space.
So, resolution and basic codecs.
Do I need 4k or 1080? It seems 4k is the assumed as standard these days for capture producing roughly an 8mp image per frame, 1080 is a lowly 2mp file, so out of date. The common logic though is, you cannot see the difference on most screens or even if you can, it is only evident on direct comparison, so future proofing aside, which is often not a real world thing, 1080 is usually what people need even if it is not what they think they need.
The key seems to be quality, not quantity of quality. Very well recorded 1080 footage can go up, oversized recordings will likely end up going down in size for storage etc and poorly realised 4k is not better than poorly shot 1080.
1080 looks to be the smart move for my work, opening up more options at the camera end and heavily reducing file sizes and media stress (4k B-RAW at 5:1 compression is 81mbs needing a 560mbs write card or SSD, 1080 is 20mbs, so a 160mbs card is fine). The cost in cards alone is enticing. Can quality from other elements make up for the resolution difference?
Codecs.
As well as resolution, codecs determine both the capture, processing and output quality and often the options available to you. MP4 is easy to use, small, very versatile, but limited. Apple’s .Mov is better and the most common serious codec for in camera use, but is not fully supported on all platforms once exported.
Codecs are often overlooked, but as we move up the chain, their importance becomes obvious.
I have used .Mov without hesitation, but may also export an MP4 version of my work if I am not sure where it will be used.
Colour profiles.
Unless you are using RAW, “baked in” picture profiles are more or less are applied starting with generic Natural, Standard, Portrait, Vibrant etc. These are camera processed profiles, designed to be shot and used as is with minimal processing.
In modern video empowered hybrid cameras, there are often video specific profiles like Cine-V or Like Rec 709. These react the same, just look more “cinematically” aligned. A common trick is to reduce contrast within these to simulate a “flatter” profile (see below).
They allow you to apply settings and features only the camera can apply like dynamic range boost and sharpness or saturation etc, but are quite inflexible in post and depending on the camera, usually have less dynamic range than un-processed profiles.
Bottom of the food chain, 10bit/422/.Mov granted, but Standard profile, Long-GOP, 1080. Fine if you get it right (which I did not loosing the shadows completely, but it was not a test of exposure). To get the file above right, I would need to do “front of house” fixes like a reflector or light.
How do they effect quality?
If you have a controlled situation, need to output your footage directly or super fast and trust your processes, these profiles are fine, because at the end of the day, much like camera jpeg’s they are where you end up anyway.
The reality is, they do not effect potential maximum quality overly (the codec does), but fall apart quickly if heavily processed.
Elements such as white balance and exposure are critical to get right because they are hard to adjust in post processing, so control at the point of capture is important.
Next step, editing profiles.
After these come “flatter” or also more video specific processing profiles which include Cine-D, HLG, Flat etc with reduced contrast, saturation and sometimes sharpness to allow for more aggressive colour grading in post and wider dynamic range.
If your subject may require some post processing, but is still in a reasonably controlled space especially with controlled dynamic range, then quite often this is enough.
Flat profile on a Lumix is my “work horse” profile and plenty for most basic shooting. I will use this with Long-GOP for most interviews etc, switching to All-i if I have it for moving subjects. For added longevity, the BMVA can be used with the non-upgraded S5II in ProRes codec (LT probably), but this is overkill quality wise, just an enabler. The advantage of Flat for me is it has a little wriggle room, but still behaves like a standard profile type allowing me to apply idynamic range boost.
The king of internal profiles.
Becoming more common and often seen as the high water mark for most videographer’s needs are internal LOG profiles, in fact often the move to LOG is seen as the step from amateur to pro video production.
Flatter even than the flat profiles above, giving us a RAW-like look, these can reward the user with quite powerful results for a little more effort, but are generally too flat to be used out of camera.
For some, LOG is not worth the effort or trade-off (it shows RAW realities like more noise etc), but the support network for LOG is substantial, so sometimes it can actually be easier to use.
Each brand has their own LOG or light-LOG profiles and they tend to remain quite consistent, so processing software can usually apply pre-set LUT’s (look-up tables of pre-set settings) to them by brand as a start in their processing chain.
LUT’s are often seen as the beginning and the end of LOG file processing, but should be seen as a step only, especially LOG to 709 conversions.
At some point, most serious videographers will end up using LOG profiles occasionally or maybe even always as they allow for both the maximum post processing flexibility and the most support of out of camera footage.
When cameras are touted as having “x” dynamic range, it will be with their LOG profile used, usually capturing more highlight range.
If your footage is important, difficult to capture due to mixed or strong lighting, needs to be delivered in a post-processable form, or you want to push it hard in processing, then LOG is probably the way to go.
Out of camera or full cinema camera internal capture.
Once LOG has been mastered in .Mov codec, there may be a desire to go “full noise” and seek out the codecs the professionals use. With these come added processing difficulty, often larger files (but not always) and usually with mirrorless cameras they require some sort of off-board recording device.
The Ninja-V or Black Magic Video Assist units for example will allow compatible mirrorless cameras to record in ProRes or various “RAW” formats as well as some other formats I am not familiar with.
A still taken from some 1080 V-Log-L ProRes 422 footage from my S5. Standard 422 is the second from best PR codec and considerably more powerful than .Mov. This footage was under exposed, but recovered nicely. Non RAW ProRes is probably the most a videographer would need as a day to day codec. It gives more quality, using the camera’s internal picture profiles and has the benefits of processing noise and sharpening internally,
Apple ProRes comes in several types, from very light Proxy to 422/HQ and are All-i formats, but still responding to colour profiles*. ProRes 422 HQ is the maximum quality most cameras can provide short of RAW and is an industry standard. According to Black Magic, B-Raw 3:1 compression and ProRes HQ are similar sizes (B-RAW at 33mbs, PR at 27.5), so the choice really comes down to a more flexible file or a more ready to use out of camera codec.
Even though they are not true RAW file type in the still photography sense, these still offer vastly superior control of white balance, dynamic range, exposure and even ISO in post processing.
At the bottom end you still have theoretically superior capture to most internal codecs, at the top end you max out what compressed internal 422/10bit can record.
File sizes are huge and hardware needs to be capable of handling that. ProRes is an Apple product, so for best hardware value, Mac’s are often a smarter move.
Some more S5 1080/LOG/PR/422 footage. This grades very nicely and has good baseline quality.
RAW, which is usually pseudo-RAW, like B-Raw or ProRes RAW is for most of us and the movie making industry the holy grail for capture and processing using 12 to 16 bit colour depth.
B-Raw has various compression rates, but even 12:1 seems to be better than the equivalent ProRes and a lighter load on a computer. Big files to record, RAW files are actually easier for a computer to process because the computer does not have to “unpack” the compressed file, it just has to deal with the sheer bulk.
ProRes Raw is even more “pure” using an alpha channel, so it is 4444, not 422 colour depth and 16 bit. This is the real deal, but massive and a huge strain on hardware.
For me?
My S5II is set up for endurance and static recording being fan cooled and V-mount powered. In 1080/Long-GOP/10 bit/422 in Flat or LOG it is enough and can be handled by pretty pedestrian V30 cards. I can record out to my Video Assist to ProRes and an SSD, but that would be an extreme case for that camera.
Some S5II and Lumix 35mm f1.8 talking head footage from a shoot recently. Long-GOP/1080/Flat/422/10bit is fine for this sort of thing, more is probably overkill. I shot some footage at the top of this pathway into stronger light and Flat was not wide enough DR to handle it, so I should have used a light or reflector then. Maybe B-Raw could have handled that.
The G9II is my movement camera so it needs to be flexible and fast, but also to be able to handle busy, complicated scenes. Great slo-mo options, the best stabe and AF of my kit, and importantly the ability to record All-i internally or ProRes to an SSD without any added weight or bulk make it the camera for this role.
A cage, a pair of side handles, a top handle, a weighted or shoulder rig and I am set (no screen as I want to use it’s touch screen focussing, but maybe a phone as screen if I can get that sorted). I could use the big rig from the S5II, but it is not that camera. In a nutshell, the easiest path to better quality, with minimal rigging.
If I could have only one camera, this with the RigidPro rig and BMVA is my power house, but too much for most things.
The new (old) GH5s is a bit of fun and a bit of a sleeper. Behaving like the love child of the G9II and S5 mk1 producing a pseudo Black Magic Pocket 4k, it has reliable AF, excellent low light performance, All-i internal recording, full compatibility with the BMVA for B-Raw or ProRes and no stabiliser. I will use this as a static cam or a traditional weight assisted hand held rig with cine lenses (which it was bought for).
I bought this over the BMPCC4k or BGH-1 for its more practical operability and compatibility with my existing gear, better low light performance, but otherwise much the same camera with a BMVA mounted with better battery life.
As a minor but real consideration, it could simply be used as a stills cam at a pinch, my “let’s drop this video crap” release switch**.
The S5 mk1 is a very capable video cam, in some ways better than the S5II, but at this point I will press it into service as a stills hybrid/reserve camera like the G9.I’s. I just have too many other options available and as a stills cam, it is basically a big G9.
If I go down the B-Raw path, this will switch with the S5II (I cannot use it in RAW without a paid upgrade) as my stills hybrid.
Another of the shots lifted from ProRes footage taken with the S5.1.
So;
422/10bit is not negotiable. Less here does much harm for little gain. All my cams can handle this, so I will accept the gift as offered.
1080/Flat is a solid base and grades easily. I have four cams that can handle this, several others with similar profiles (Cine-D, Natural with contrast reduced).
Long-GOP with the S5II empowered by a fan and V-Mount battery, with the GH5s or G9II (NP batts) as support, will handle interviews and static endurance work. It is efficient, though a tough ask for my computers. If I keep it linear and do not mess with it too much, there should not be an issue.
All-i is for movement, slo-mo and similar on the GH5s or G9II. This will not be long connected reels so the extra depth or shorter endurance will not be harmful. Potentially ProRes (1080 HQ, 422 or LT?) may replace this as it is actually better while still being All-i file size and I can use it with the same number of cams or potentially more if I add a BMVA 3g (the non B-Raw capable Video Assist).
4k will only be employed when (1) requested specifically or (2) for zooming in post. The fact that I can use 1080 3:1 B-Raw or ProRes HQ for similar strain on my system as 4k 12:1 B-Raw or ProRes Proxy is compelling.
ProRes/LOG will be used for special projects and very important stuff, which is currently limited to 2 cams, three if I get a 3G BMVA. The extra quality of LOG and ProRes would be noticeable on close inspection, but that will rarely be needed.
or
B-Raw at 5:1 or 8:1 if maximum post processing is required, but this is limited to one cam unless I get another BMVA 12G. I am torn here, but need to do more of each to see. ProRes is so much more accessible, B-Raw potentially so powerful and system friendly I may use it more and more and 8:1 is much smaller than PR HQ.
*A question I need answering is there a difference between a Standard profile in ProRes as compared to LOG or a flat profile in .Mov?
**I only have a few things that would be considered truly useless if I moved out of video, something I like to keep in mind.