RED wishlist

someone said that the ability to control via laptop would be good, I agree to a point. I would like to see that ability but would like to also see a compact flash card slot to be used as a port for setups. a DP could have 10 or 12 cards (low memory=low cost) that have their favorite setups on them. one for a bleach bypass look, one for a kingdom of heaven look, one for a uber saturated high contrast look, one for a kodachrome look, etc. these cards could have the virtual filmstock and the virtual post processing look saved onto them and could be swaped out at a moments notice. the filmmaker could have a ton of setups saved to his computer and just put the ones that he might need on the CF cards.
 
Graeme_Nattress said:
I think the JPEG2000 format covers up to 16bit per channel, but the only software implemention I've used, in Quicktime, is 8bit.
Click to expand...

actually, according to specs, JPEG 2000 supports a gazillion of colour channels (well, 214), each with up to 38 bits.
don't know any application that uses more than 4channel (RGBA) at 16bit either though.. and as so often, quicktime is only 8bit (but at least has alpha).

but you're right, without hardware acceleration it's pretty slow, so it's not a good editing codec for low cost systems.

++ chris
 
Thanks Zem - the spec of JPEG2000 is fine, but I've not found an implementation in hardware or software that lives up to the ideals of the design. Anyone know better?

Graeme
 
mike over at hdforindies has posted quite a lot about jpeg2000 specs for digital cinema... they have quite impressive resolution, bit depth and frame rates specs, so for hardware implementation there's definitely some high end stuff out there. but then again, they have some serious cash to spend on each system.

as for software, there are a few obscure image applications (ok, photoshop too to some degree) that have implemented it, but as said most of them are only 4channels 16bit... i guess the reason it didn't take off is that storage got too cheap to worry about a few MB, and .tiff and .psd dominates the print world anyway, and provide a lot of extra features (who wants to discard all layers everytime he saves a pic?).

and as a software codec for video, it's simply too processor intensive :/
++ chris
 
btw, all this got me thinking about one of my major concerns about the red camera..
(and you're probably the right person to exchange ideas about this, since you're now in the red team ;)

to me the major problem to keep this an affordable solution is data throughput/storage...

highest quality/lowest storage would of course be to shoot raw and do the image processing later on with high quality de-bayering.
but that's 394mb/sec on a full size mysterium sensor with 12bit... simply wont be an option for most of us.

so let's say there's a high quality debayering and resampling algroithm that does 1080P in real time. we're now in RGB though, and we would at least wanna keep 10bit, which means uncompressed that's still 178MB/sec. again, i can't think of an affordable solution to store this to any portable digital mag (ok, uncompressed is overkill most of the time).

putting this through a HD-SDI (or two) is of course cool for monitoring and for studio work (direct to computer recording), but again renting a HDCAM SR recorder simply wont be an option for most of us, and rather cumbersome for location shots.. and just the cost of the tapes make me cringe ;)

so it seems to me that the key is finding an affordable capture medium, which doesnt carry huge post-production costs (ie no tape format).

the best thing i can think of atm would probably be a raid0 of some (four?) 2.5" drives. they should be small enough, provide capacity up to 400GB and be able to handle some pretty high data rates (probably over 100MB/sec over the entire range). that would mean we'd need a codec that does realtime compression of full raster 1080P with about 1:2, possibly lossless (one can dream, can't we). that would give us about 1hour of footage per mag - and those could probably be built for about 1000 bucks).
convenient storage (with backup) would run into about 500USD per hour.
(and of course 20-30MB/sec full raster codec would be really nice too, for all those who don't plan for a filmout ;)

one other option would be to use only the center part of the image (about 2K) and record this as raw, with the upshot that we could use affordable c-mount lenses.
this is also around 100MB/sec at 12 bit though.

maybe by the time the camera ships a CF card raid would be an option, a 40GB raid with 100mb/sec should be doable for 2000usd by then i guess... and give us 7mins runing time at 1080P 12bit rgb with 1:2 compression.

oh well, just some thoughts
++ chris
 
Zem, what you say is very interesting. I don't see why you need everything uncompressed though. SR is still compressed, and if that's good enough....

Graeme
 
Graeme,

i don't really need uncompressed myself, i'm just thinking that if somebody designs a camera with a 4K sensor, there should be an affordable way for to shoot at least full raster high quality 1080p (with target: filmout). and it should be nearly uncompressed as more and more films need greenscreen footage (a clever 1:5 compression would probably be acceptable though).

the problem i see is that there is no affordable industry standard high quality full raster HD codec.. D5 and HDCAM SR are tape formats, but afaik on disk you always work in either a proxy resolution or uncompressed (or is there such a thing as a HDCAM SR codec on disk? and if yes, is it available for any low cost editing suite?). HDCAM is not even full raster (and not really an editing codec either). DVCPRO HD is the most interesting codec to me because you can edit natively on low cost systems, and only bump it up on your final grading (similair to what DV did for SD) but is only 8bit and 1280x1080px.

to extend that coparison to the SD situation about 10 years ago
(i remember editing in 1:2 compression on a 30K Media100 system which could hold 1 hour of footage):

HDCAM SR and HDCAM = Digibeta
meaning, it's a tape format, protected by sony and has to be brought into the computer with expensive hardware/interface

DVCPRO HD and HDV = DV
cleverly compressed footage that still looks good, can be captured and edited with consumer hardware, but falls apart on heavy manipulation .

what we're missing is something like DVCPRO50, ie a compressed codec that is cheap to handle, virtually transparent, can be captured and edited in online resolution for the whole process (no recapturing needed for final output).. . Cineform would probably be the closest that i know of (supporting 10bit), i just feel uncomfortable that it's not available for mac, and i dont think that the support in the professional post houses is very big. so you very likely would have to record to a self-designed codec, which means generating proxy files for editing, and a lot of data management issues on longer projects (i'm sure you've been there ;)

maybe one could have the camera generate a standard "offline" resolution file at the same time (preferably DVCPRO HD 720p) so that some stuff can already be edited on the field on a laptop. or maybe you can generate a codec that hooks right into quicktime (that would be sweet)!

but my fear is that the camera will get too expensive if it's got to do all this realtime conversions/compressions, personally i'd be perfectly happy with a camera with a 2K sensor that records RAW 2K and DVCPRO HD 1080P.. (as long as it stays around 10'000usd and uses a reasonable storage system)

++ chris
 
Thanks Chris. You've really pointed out the issues with the choices of codecs, and especially how tape based codecs are not suitable in today's non-tape-based world.

Graeme
 
ok, since the topic reads "wishlist" and somebody seems to be actually reading this i thought i'll add a few thoughts about the camera design rather than all this codec talk (being a cinematographer rather than a post-production guy)..

there are several types of projects, all having different needs:

first, the indie filmmaker (which what i would count me in):
shooting unter controlled conditions, they look for a camera which has full manual control, possibly with a standard lens mount, a 35mm sized sensor. you could skip autofocus, automatic exposure, basically it sould be a film camera with a diital mag etc...
the most important thing here would be high quality and affordable! which means cheap standard recording media (atm, hard disks seem the only way), cheap monitoring (DVI instead of HD-SDI out?). since you'd probably record RAW anyway (although not at 4K, as pointed out before) you wouldnt need a 30'000 HD production monitor for color/exposure (btw, built in histogramms for exposure are a MUST!! it's standard in still photography, why not in film?).. being able to hook up a high res LCD for focus checks would be great though.
camera should be designed like an Arri SRIII or even better an Aaton XTR. heavy and sturdy. optical viewfinder would be great but probably make it too expensive.
there's not a single camera out there which is sutable for this kind of work (maybe the kinetta does once it ships)


then we have the professional filmmaker for theatre release:
to them, quality, reliability and compability with existing workflows are more important than specs and money. think dual HD-SDI, HDCAM SR on the low end, and Arrri D20 on the high-end. they rent anyway, and the crew (and actors) will be much more expensive then the camera - if there's a technical defect that delays shooting a few hours, or some data loss, it will be terribly expensive. so everything is built like a tank, even if it costs a fortune (think paraniod).

professionals for tv shows:
not much experience there, but afaik most important factor is ease of workflow... episode has sometimes to be shot in one day, and often edited in real time! sony, panasonic et all will always win in those markets since they are optimised for that sort of work.

broadcast/news/docu type of camera:
actually a full size 35mm sensor doesnt make a lot of sense for that... focusing would become an issue as well as way too much data. basically ENG form factor cameras with tape recording (HDCAM or DVCPRO HD) would be much more suitable for that, since you dont have to worry about data management. all the automatic functions (autofocus, image stabilizer etc) would be rather difficult to design i guess. plus, there's a lot of competition out there already.

typical prosumer, doing cooperate ads, weddings etc:
actually this kind of person would be best of with a HDV camera, nobody will ever notice the difference :) - but a lot of them take great pride in delivering the best possible quality (cant blame them). but again they'd need autofocus etc, and a full size sensor would be rather problematic, not to speak that nobody would have the capacity to work with and store the huge amount of data.


hmm, i'm sure i forgot some applications, personally i'd think the indie filmmarket would benefit the most from such a camera, but only if you can find a way to make it affordable and find a solution to the high data rates. alternitavely, you could try to compete with the D20, but then stuff like optical viewfinder etc would probably throw it out of reach for most on the people on this board.

well, just a few ideas
++ chris
 
In a year or so I would expect a 150 gb 10,000 rpm Sata drive to cost about 200 bucks. Now if there was a way to raid 0 four of them at about 210MB/s you have enough for uncompressed 1080p (10bit). At 800 bucks for a 'mag' of 600 GB (almost an hour). You have yourself a pretty nice storage system. The problem though: 3.5" hard drives aren't designed to operate while moving. They're also not exactly energy efficient, you would probably need to run an extension cord to your RED. And even then you're looking at the whole shock issue, although I would assume this would be remote and in a backpack of sorts.

I guess what I'm saying is... wishlist: Functional SATA Raid brick.
 
Able to save ALL user configurable settings (not just "film looks") to say widely available compact flash or SD card?

Useful for storing your own scenefiles for different shooting styles, matching multiple cameras, minimising setup time with rental cameras, reshoots etc.

Also able to print out settings, or save to industry standard metadata files.
 
Bit pie in the sky but how about - Advanced interval/timelapse recording options? Eg. ability to change iris/gain/shutter/focus/zoom settings over any recording period up to several days or unlimited? Either straight ramps, or if linked to a laptop program, even tweakable curves for individual frames or between multiple in/out points?

Set the tape length and recording period and it automatically calculates the required interval between frames.

Some sort of Output which could be used to trigger external kit like motion control camera mount, flash lights, etc.

Some sort of Input which can be used to trigger the camera to record a frame, or record video continuously, or for a user set period? Useful for wildlife filming?
 
I guess this kind of detailed control would be better implemented on a laptop that can control the camera? To be able to program this direct into a camera would be painful indeed, whereas if there's an API for the laptop, you could, I guess, write like a shell script to make it do your bidding? Thoughts?

Graeme
 
Graeme_Nattress said:
I guess this kind of detailed control would be better implemented on a laptop that can control the camera? To be able to program this direct into a camera would be painful indeed, whereas if there's an API for the laptop, you could, I guess, write like a shell script to make it do your bidding? Thoughts?

Graeme
Click to expand...

Absolutely. But rather than leave it to a third party, it would be great if there was a general purpose GUI written by Red that could access camera parameters, and do all this? But I guess this is getting too specialised and wouldn't be of use to the majority of users, so an API would also be a good idea.
 
Handle strong enough to mount on a Steadicam with a low mode handle clamp.

Handle removeable/replaceable - offers more mounting options, lower profile, replaceable if damaged.

Baseplate stronger than the DVX100's, but also replaceable rather than part of the casting in case it does get broken. Perhaps an extra 3/8" mounting screw socket in addition to the usual 1/4" one, further away from the tripod locating pin. This would offer more security and less "twistability" on a professional quick release plate.
 
I have a few ideas that would be useful to have in a 21st century camera.

This camera could do for video/film what digidesign protools did for the audio industry. Why not everyone building their own workflow around a common platform? This divide between the pros and independents is so 20th century. And in that light how about:
  1. All camera functions and operations controllable wirelessly through your PDA.
  2. Automatic shot logging to that same PDA.
  3. Camera functions upgradable in software as plugins.
  4. Camera "software plugin" functions available to third party developers with an extensible API
  5. Camera generated meta-data (i.e. GPS info) that could be used for aiding camera matchmoving in 3D applications like Maya. Or future lenses that provide depth information for real-time compositing.
  6. Many camera functions adjustable remotely to allow for automation (i.e. Operator settable parameters which can be animated over time.) such as user defined changes in frame rate or exposure or operation remotely. Or how about ramping various operations with a user defined curve. Such as ease-in or out when animating a color balance? This would also aid in using the camera in inaccessable environments such as underwater housings.
  7. How about several image sensor options and the ability to plugin a new sensor at any time. This would also enable spare image sensors to be carried as safety backups in remote locations. Or image sensors tailored towards different projects. No different than alternate lenses or microphones for varying projects and budgets. Not everyone will need 4k output.
  8. It would be very cool to be able to customize the monitoring functions on a per shot basis. How about real-time superimposition of a histogram in the operator’s monitor.
  9. Since focusing is a huge issue, how about a monitor output that analyzes the picture and automatically shows areas of hard contrast to aid in focusing.. something like a zebra display but designed for aiding focus. Something that can indicate when maximum focus is achieved in a given area of the frame or some form of display which isolates edges and gives feedback on how sharp the focus is on those edges. And why not allow 3rd parties to develop additional monitoring types?
Basically I think the key is to allow the users to customize the camera for their own needs and allow their inventiveness to drive the cameras’ evolution.

Cheers,
Pete
 
I'd like a marker, just like the HVXs, but with a smaller target zone. A true built in spotmeter.

And I'd like if the camera had a program that could calculate DOF, if the lens trade some info with the camera, such as focus distance, aperture and focal lenght then the depht of field could appear in a corner of the viewfinder... (how cool would that be!) You can change the COC in a menu. That'd be a revolution!
 
On board laser range finder? Actors would of course have to have some kind of protective face shield when in use. Great for wilderness filming etc.
 
WesVasher said:
Buy one, attach to cage, now on board, done.
Click to expand...

But I want to point my laser to other places! If all I wanted was to know the distance of stuff in the center of the screen, I'd use autofocus! :)
 
You must log in or register to reply here.
Back
Top