Concerns and Questions about 8K

[Neil Dearman]

Everyone on this board knowns I bleed/sleep/eat RED Products but I have concerns about the 8K Cameras, only 2 btw...

Storage/Data/RED MAGS:
On Reviews with the StormTroopers that using a 1TB mini mag shooting 8K will only an hour of footage (that concerns me because I shoot a ton of music videos/commercials) and most productions only have 2 4TB RAID Drives. An average with my 2 weapons we go 8 cards each and almost max the drives out when wrapped shooting 6K. Also RED MAG 1TB aren't cheap I usually use 120s rarely my 512 (cause they take a long time to transfer). So My DIT's won't really be happy transferring 1TB/Plus the Producers will also not a fan of that much footage on one MAG.

Won't if I don't want to shoot 8K?:
Talked to a lot of my DP's/Clients about 8K and they say they don't need or want they cause of the Workflow/turn around times. And if i shoot 7k/6k I will punch in so making a 18mm to a 35mm or 35mm to 50mm, (unless there is a way to make it where you don't punch in which would be epic).

Thats really my major concerns and the big one is Data/Storage on big shoots. If anyone can chime in that would be epic about my issues thanks guys looking forward to the new Sensor!!!

 

[Anton Shavlik]

Just shoot 12:1 compression unless you really need all 8k of resolution or are worried about noise. 12:1 at 8k is roughly 6:1 6k data rates, and as we've seen with tests on the dragon, it's better to go higher res with more compression than the other way around.

 

[Nathan DuMoulin]

Your concerns make me curious about how Epic-W will do on the rental market versus my existing Scarlet-W. With so many productions not wanting to deal with 7-8K, I wonder if it would actually be detrimental for me to upgrade?

I think I've switched my mind back and forth about 10 times now.

 

[Tim Daust]

Just shoot 12:1 compression unless you really need all 8k of resolution or are worried about noise. 12:1 at 8k is roughly 6:1 6k data rates, and as we've seen with tests on the dragon, it's better to go higher res with more compression than the other way around.

Phil Hollands FORGED was shot 12:1 and it looks great.

I think there is a rumor about a 6k or 4k r3d wrapper for 8k FF.. not sure where that rumor started but that would be sweet! not holding my breath tho

The other thing to consider about 8k is computer power for rendering/playback, going back to the day of proxy workflow

 

[Akin A]

I told myself that Epic Dragon 6K was good enough for me for the next couple years, but that temporary Epic-W Helium for $20K deal has me wondering if I will regret not taking it and seeling my Epic Dragon. I'd really like to take 8K stills, get cleaner shadows, an alleged nicer looking 4K downscale for delivery, and just to have the latest shit, lol.

The additional cost of a base I/O and likely a 1TB drive (512GB seems like it'll fill too fast) and more storage on my computer is telling me to chill out, for now.
I'm hoping somebody will say something that makes me stop considering it, and stop looking at that Update Thread, lol.

 

[Björn Benckert]

If you Finnish in 2k, HD or such, or even 4k then 8k capture make no difference in data rate, the mags and data rates max out on the same number and as written above resolution beats compression, Better shoot twice the since at double the compression, as it improves picture quality but does not make the files any heavier.

So producers and other that does not want more K's when shooting are struggling with poor knowledge and make ill based assumptions.

1Tb mags are a risk according to the "all eggs in one basket" theory, but swapping smaller mags and cycle them in higher rotation also has it's risks. Simply the human error comes to play and I find it to be a greater chance that a DIT format the wrong card than the camera creates an error. Even if there has been some issues recently.

So I would say no need for more or bigger media. If you recorded on lowest compresson at all times before on weapon, then simply the same practice will not give more data when recoding in 8k as long as red does not release some faster mags.

And the post side of things does not really change much as you can edit r3ds with up to 1/16 proxy mode and even if your render with full debayer will take longer it's a question of minutes when making music videos.

And further more you also have prores proxy recording to lean against in those discussions.

One thing that I think would help this even further would be if red made 8/1 debayer possible in redcine renders then also transcoding proxies would even be faster than 6K at 4/1 debater for example.

 

[Tim Daust]

while Neil's concerns are legit I think that this is similar to when we went to 4k from 1080p or even from SD to HD period. It takes forever for production to catch up with the demand for the acquisition to be so high. right now most places still take 1080p deliverables. When a majority of deliverable requirements move to 4k (which NETFLIX and AMAZON already are pushing towards) you will see all the productions rushing to 8k standard acquisition. Its always like this when we shift, but the key to remember is staying up to date with technology is the only way to advance the industry forward period.

Personally I think that is what Jarred & the RED family are doing here with taking everyone by storm with the 8k push. I mean look at Canon's latest 8k prototype.. for real.. a prototype. get real.

RED is paving the way and the train has departed.

5 years from now we will look back and laugh at how bad 4k was .

just my 2 pennies

 

[Hans von Sonntag]

...and as written above resolution beats compression,...

In my experience it depends on the detail, complexity of the shot how much compression affects picture quality in an problematic way. Remember the tree of death which broke R3D compression all the time? I think that the math resolution divided by compression does not necessarily works as one might think.

---

In the German rental market Weapon is a rarely found beast, 95% is Arri or legacy Red Dragon like my Epic which rents out significantly cheaper than an Alex Mini, a low budget choice in many cases. This will eventually change when Epic-W will hit the market. The new Helium sensor in a S-35mm 3-perf format with 8k gives an interesting edge compared to Alexa. In TVC land re-framing is often wanted - the extra Ks make this an even better option than we now have with 5KHD, the format of choice when PL lenses are uses.

Epic-W will be an interesting option for me. The only question I have: does my current PL glass resolve 8K? I own a set op Super Speeds and a short Optimo. My uneducated guess for now is that the Super Speeds won't resolve beyond 6K the extra resolution. Probably I have to invest into new glass for 8K resolving sensors.

--

Data rate is an issue. Apart from the costs of high capacity Red SSDs the technological edge is always challenging computing power. But the good news is that debayering 8K half will be effectively like 6K full. So if 4K is the target format a half debayer will be sufficient in most cases and won't take much longer than todays 5KHD from the Dragon sensor. 8K full will be only necessary if the best possible quality is needed, e.g. green screen footage or magnifying is planned.

Until now I never had a project that required more resolution than 2k. From this perspective my Epic Dragon is still one of the best options in the market. Since I don't rent out my camera I can happily leave things like they are. But time is moving and 4k knocks on the door. I recently invested in a very fast linux based ADSK Flame that makes 4K not only possible but almost something like convenient. Obviously the camera must follow this route at some point. Epic-W will be an interesting option in a year or so. But the competition doesn't sleep. NAB 2017 will be very interesting in this regard.

Hans

 

[Tommaso Alvisi]

But the competition doesn't sleep. NAB 2017 will be very interesting in this regard.

I don't personally care much about 8K but downsampling is always good...in camera 4K from an 8K acquisition is the sweet spot I think...

BTW I've seen Canon 8K prototype @ PhotoKina.

Apparently they give you a small truck for every pre-order. Ha!

 

[Nick Morrison]

Data sizes are def a concern, as they always are when shooting high resolution.

At same FOV, Helium would be double the data rate of Dragon (ie 7K vs 5K).

Shooting on 2 Heliums at 7K, would be equivalent data to 4 Dragons at 5K.

Shooting 3 Heliums at 7K, would be equivalent data to 6 Dragons at 5K.

However...you would have the benefit of much higher resolution!

Everyone needs to figure what's most convenient/productive for them!

 

[Michael Tiemann]

while Neil's concerns are legit I think that this is similar to when we went to 4k from 1080p or even from SD to HD period. It takes forever for production to catch up with the demand for the acquisition to be so high. right now most places still take 1080p deliverables. When a majority of deliverable requirements move to 4k (which NETFLIX and AMAZON already are pushing towards) you will see all the productions rushing to 8k standard acquisition. Its always like this when we shift, but the key to remember is staying up to date with technology is the only way to advance the industry forward period.

Personally I think that is what Jarred & the RED family are doing here with taking everyone by storm with the 8k push. I mean look at Canon's latest 8k prototype.. for real.. a prototype. get real.

RED is paving the way and the train has departed.

5 years from now we will look back and laugh at how bad 4k was .

just my 2 pennies

^ This. Except I don't think we'll laugh at how bad 4k was, at least not at those who used it appropriately. There was a time when 16mm film was a crappy choice for doing low-budget VistaVision. But today, 16mm is a valid aesthetic choice for those who want that look. When used appropriately it's not a crappy choice, it's a valid choice.

In the digital video world, 1080p can certainly do some jobs well. 4K can do those jobs well, and more jobs even better. And I salute those whose pioneering spirit compel them to find those new 4K opportunities and bring them to the screen (large or small). 8K brings yet another range of technical ability to the filmmaker, and in the next few years we will learn what are the new territories that it opens up. It might be more on the capture side (easier reframing revolutionizes shooting techniques). It might be more on the quality side (where higher resolution/higher compression really does beat lower resolution/lower compression). But one thing's for sure: the whole workflow from capture to finished product is feeling as cramped against the 8K performance wall as 4K was 10 years ago. We need more powerful CPUs, GPUs, higher bandwidth network infrastructure, more dense and cheaper SSD, and computer hardware vendors willing to really throw open the doors to higher performance workstations. Apple's Mac Pro is how old now?

TL;DR: one doesn't simply walk into 8K.

 

[Phil Holland]

I've been fortunate to be an early bird on the 8K front and can share a few thoughts on working with 8K.

As of now a 24fps 8K FF project with a 5:1 REDCODE RAW Compression Ratio gives you about one hour of role time on a 1TB RED MINI-MAG. As mentioned earlier, you can change your REDCODE RAW Compression Ratio to say 12:1 and you'll have 3 benefits from that:

- Longer possible role time on a single media card
- Easier to handle post related bandwidth (i.e. NLE, Color Suites, etc.)
- Smaller Storage Footprint

That said, I mostly role at 8K FF 5:1 for the quality level. I'm also finishing my work at 8K and 4K now however, my needs may be different from yours for sure. Heck I wouldn't mind 8K FF at 2:1 for that matter


Media Thoughts. I think the 512GB MINI-MAG is a pretty sweet spot for most shooters. When on commercial shoots often DITs want you to send them mags at around 50% capacity to help speed that up. That's not exactly how I hand my sets, but it's pretty common. Personally I roll 1TB MINI-MAGs and 512GB MIN-MAGs. A 1TB media card takes about 30-37 minutes to offload on a properly configured sytem over USB 3.0. Jarred teased a USB Type-C RED Station which will likely increase offload performance. I've tested pure read speeds off of MINI-MAGs and they can indeed read faster than what USB 3.0 nets you currently, so that's going to be an improvement.

Personally I wouldn't mind 2TB MINI-MAGs or hell even 4TB MINI-MAGs. I don't have any "trust issues" by having so much footage on one mag. I haven't since I've been rolling the 1TB media cards since they've been released. Heck I'm still using the same cards I was using last year with no issues.

Don't know about DITs using only 4TB RAID drives for offload systems. I don't think I've seen one with anything less than 8TB in the last few years, but there's certainly different markets and budget levels.


To put many things in perspective on this current seven figure shoot I am serving as Director, Director of Photography, DIT, and Colorist before the footage is sent to post. This is my project and the style of the shoot allows for this chaos, which I'm happy to have in this case. I'm working out of hotel rooms around the globe on a very powerful laptop and using USB 3.0 RAIDs to offload to and I also have this big and slow 10TB bulk storage for triple redundancy on this project. A copy goes to studio, B copy stays with me and the Producer, C copy stays with me. If the drives don't get back to LA in good shape (which hasn't happened yet, thanks FedEx!) I generate a new copy to send their way.


One thing I will say for those serious about working with 8K footage on the post side is you'll want a GPU with a good amount of RAM on it. Especially if you're a plugin user. An 8K frame once uncompressed can cause fun issues on lower end GPUs. If you're using any of the recent and last generation mid to higher end NVIDIA cards that's a good place to start. Multiple GPUs can help out in a big way. A dedicated RED ROCKET-X card is still at the moment the fastest option for debayer and dailies, though the gap is closing year after year. So much so that after PCIe 4.0 drops I would wonder what the future holds for the Rocket concept, but that's over 2 years away from hitting the market as of now. Perhaps closer to 3+ years if mobo manufacturers are slow to employ/deploy. A few other things will be faster by then and we'll be having a rather different conversation at that time.

Transcodes otherwise take time whether you've got a fast or not-so-fast system. The modern DSMC2 cameras that support dual record to REDCODE RAW and an Intermediate Codec (ProRes and Avid DNxHR/HD at the moment) should really be explored if that's you're focus and game plan. Dailies are made immediately in camera and during offload can be properly organized very quickly.


8K won't be the answer for everybody at every budget level, but it's certainly manageable in a variety of situations. The post house dropped their jaws when we started filming in July at the idea of working with 8K material, but have since settled down to "oh this all works pretty well". The key is "Don't Panic" and look at how to actually achieve those goals. For some larger facilities and productions it's a harder conversation as 4K is still a challenge. But many are now exploring 4K workflows and not so focused on the word "can't".

 

[Louis Maddalena]

Phil, do you happen to know if the RR-X will work with Helium? With 8k, I'm thinking of putting one in my system but if its fixed to Dragon like the original was fixed to MX then buying one now would be a waste of money since I hope to have a helium in my camera as soon as they are shipping to Carbon owners.

 

[Austin St. John]

Thanks Phil! Always enjoy the knowledge bombs you drop. Can't wait until you get your hands on Helium and can tell us your thoughts

 

[Brendan H. Banks]

Thanks Phil! Very insightful and helpful as a person in the S-W/E-W dilemma.

 

[Ulf Krentz]

I would love to see an internal 8k to 4K downsizing / Pixel binning option. I know storage is getting cheaper all the time but it still is a massive amount of data even at 6K.

 

[Tim Daust]

I've been fortunate to be an early bird on the 8K front and can share a few thoughts on working with 8K.

You sir are a champion.

 

[Marc Wielage]

That said, I mostly role at 8K FF 5:1 for the quality level. I'm also finishing my work at 8K and 4K now however, my needs may be different from yours for sure. Heck I wouldn't mind 8K FF at 2:1 for that matter

I'm curious who in LA can handle 8K, particularly for an end-to-end workflow. I question the need for it purely from a display device problem: there's not a lot on which to watch 8K.

Michael Cioni said back at Cinegear in June that he expected most clients would shoot in 8K and then transcode to 4K and use the 4Ks most of the time, except possibly for green screen VFX and blow-ups. To me, that makes a lot of sense because your options for post and delivery are a lot greater in 4K.

 

[Mike Nagel]

Shooting 8K but having to fall back to much higher compression because of not enough available bandwidth is idiotic... Way too much ado about nothing... And then u gotta push 8K (low quality) pixels in post... Still 8K pixels... Again, with almost no benefit...

and as Hans said, the compression / res analogy is def not straight forward...

U need to shoot low compression... Otherwise u're just fooling urself. And the immediate 8K to 4K down sample is amusing. Again, u're fooling urself.

 

[Michael J Brennan]

U need to shoot low compression... Otherwise u're just fooling urself. And the immediate 8K to 4K down sample is amusing. Again, u're fooling urself.

For a 4k deliverable, an in-camera down sample from 8k to 4k is a good and viable option, compared to doing it in post if the goal is to reduce compression and reduce location data loads.
The trade off (may be) ability to simultaneously record raw, we'll see.

It looks like supersampling/downsampling from Xk to Prosres 4k in-camera will be available from a few manufacturers.
Perhaps there is something about prores that makes high quality supersampling a no brainer to implement?

Off the wall question, could raw be supersampled and maintained as raw? methinks this is a whole lot harder than doing it within pro-res??

Mike Brennan