Future improvements...still possible?

[Joel Kaye]

That may also mean 4 times as much processing time in post, and a lot of flack from my editor and producer, who may prefer that we shoot with a camera that has a chip which is 4 times smaller and doesn't look half as good.

You don't have to work at 4K resolution. A lot of workflow options will come, but you can convert your 4k to 1080P and go from there. Or you can simply convert the 2K stream to 1080P for a faster conversion.

Yeah, there's a little more raw storage required for the master files... but I think they had to shoot to 4k in order to deliver 35mm DOF. So until someone invents a 35mm DOF camera that shoots 1080P this is what we've got.

You can put a 35mm adapter on a Sony EX1 and work with 35mm DOF at 1080P. So there are options out there for you.

 

[Maz Mawlawi]

Jim,

Thanks for the answers. Just to clarify one more thing, that also means we will not have a 1080P output option from the HD-SDI? Again not a complain, just a question to know how to plan...

Thanks



P.S. The problem with the EX1 is its crappy long GOP format.....why...oh why did Sony stick with this crappy codec?.... ;-)

 

[chuckt]

I still do not understand why anyone would use RED to shoot at 1080p.
Is it for using 35 mm lenses?
Buy a cheap sony ex1 for beautiful 1080p ($6.5 K).
Add a 35 mm lens adaptor for 35mm dof ($15 K).
Why torture yourself with RED for 1080p ($45 K+)?

A cheap re-conditioned Arri 535 may be the best choice.
(easy production and post production work flow; true 35mm lenses; frame rate of your choice; true film look; can be scanned to 1080p/2K/4k.) ($40 K+)

(PS: I am not knocking RED. RED is a good 4K camera with a cumbersome production & post-production work flow. )

(PS: I am not connected with Sony / Arri.)

 

[Derek Wiesehahn]

I still do not understand why anyone should use RED to shoot at 1080p.
Is it for using 35 mm lenses?
Buy a cheap sony ex1 for beautiful 1080p ($6500).
Add a 35 mm lens adaptor for 35mm dof ($15000).
Why torture yourself with RED for 1080p ($45000+)?

A 35mm lens adaptor on a 1/2" chip camera is not the same thing as shooting with a camera with a full sized chip. For one thing, there is a great deal of optical degradation through those lens adaptors, and also, a 1/2" chip is not going to have the same sensitivity as a full sized chip.

One more thought:

I have shot many thousands of feet of 35mm film which was telecined down to NTSC. That has always been the convention for commercials. Why did the industry waste all of that money when we could have just shot DV or Beta? We were just going to throw away all of that extra resolution anyway.

The point is, it just looks better to shoot with a full sized, high resolution, wide tonal range medium and pull the reduced resolution you need out of it. It would be great to be able to do that in camera with the Red One or the next generation of Red cameras.

 

[Zach Hilton]

It appears that's the case. This is this first time I've seen Jim state that 1080/60p scaled is not possible... maybe I just missed it before. I've seen the "60p is the max framerate we can get from the full sensor!" quote, but currently I can't get 60p from my full sensor.

Are you serious? Jim and crew have repeatedly said that 1080p 60fps is never going to happen. Also tied to that, that 1080 out whilst shooting 4k also is never going to happen. And the only way you were going to get 60 fps @ 4k was through the RAW port....and then your not recording 28 - 36 MB/s but 1 GB/s which would require slightly different hardware than a CF card or a Red drive.

See:
http://www.reduser.net/forum/showthread.php?p=153297#post153297
http://www.reduser.net/forum/showthread.php?p=152900#post152900

 

[Joel Kaye]

A 35mm lens adaptor on a 1/2" chip camera is not the same thing as shooting with a camera with a full sized chip. For one thing, there is a great deal of optical degradation through those lens adaptors, and also, a 1/2" chip is not going to have the same sensitivity as a full sized chip.

I agree with your post.

I'm not sure what you mean by sensitivity though. I've shot the EX1 alongside a RED and the EX1 is a faster camera by my estimation.

But as to quality - yeah I really think RED, even downrezzed is going to give a superior asthetic to the EX1. But I think the EX1 with adapter is still a great choice if you can't do RED.

 

[Jeff Brue]

Err,
Not quite seeing what's wrong with at least the concept of the workflow. Ultimately it comes down to adoption of data workflow, instead of video workflow. Thats the real issue, belive me I know, I've been battling this for two years...

Client: But we're shooting HD! We get a piece of tape..
Me:Not with us you don't, we provide editorial files to support whatever system you want to edit on and then DPX's to color...
Client: But we want to do a tape to tape !?
Me: But its cheaper and better if you do it this way.
Client: My turn around time doesn't support that
Me: ... really I'm faster than a tape to tape.
Client: But I want all the footage on a video tape color graded

Sadly I still don't have an answer to that last one for myself but for Red its scratch on a skulltrail machine. Believe it or not you're going to have the same problems with an EX1 as with a red.

 

[Häakon]

I still do not understand why anyone would use RED to shoot at 1080p.

...aye yay yay.

For the umpteenth time, the reason is (was) to get full-frame readout at faster than standard framerates. Period. For some reason, it's not sinking in with some people that there is still currently no way to do this!! If you want to overcrank at all, you are forced to lose resolution and FOV to get it. Period. For some, that may be acceptable. Many of us don't want to give up the full sensor, however. We want to use our 35mm lenses as they were intended. What happened to the optical port that allowed us to do this (albeit uncompressed)? What happened to the scaled options that were once on the feature chart? Things change, I totally get it. Maybe things were just too complex for the system to handle and they overestimated what the processor was capable of. But it wasn't until today (that I heard anyway) that a scaled 1080 solution had been completely stricken from the plans. Judging from many of the posts here, I wasn't the only one who would've found this useful.

Strongbad_2Z, the posts you referenced show that Jim had made it clear he was not going to persue a *2K* scaled option. He has several times mentioned that they will still move forward with a 1080 one at some point - and since the old feature chart showed 1080 RGB scaled - I assumed he was talking about a scaled option, not a windowed one. Obviously that was a poor assumption. I concede that point entirely.

That being said, I agree completely with SF Geek. There's no point in wasting time or resources creating a windowed 1080 mode if no one wants it. 2K windowed already throws away 75% of the sensor and the resolution hit is quite severe. If the absolute priority of broadcast shooters is getting a 1080 file from the camera and are willing to give up depth of field, field of view, and the monster resolution benefits that the 4K sensor offers, then I personally think they're using the wrong camera.

 

[hybris]

I don't think Red should bother with 1080 cropped.
I don't think anyone here is asking for that and I was one who was really pushing for 1080 scaled.

QFT

The only thing that makes Red or Scarlet interesting to me, is the possibility of shooting 1080p in full format 35mm with my Nikon lenses.

If it's impossible with the current technology, then please make this option available on Scarlet.
(Pretty please with sugar on top?)

 

[Andrew M.]

I am reading this thread and it looks like someone have to come up with educational 3 minutes clip explaining this 1080p on RED.

I see few misconceptions:

1)That shooting 1080p full frame will be faster than 4K frame so 120fps will be probably possible.

2)That storing 1080p on the disk acquired from full sensor will take less space at much higher quality of image then 4K

3)That shooting 4Kfor 1080p workflow is space and time consuming like 4K end to end workflow.

I understand that people that do not have the camera are confused, but people with the camera?

This shows that even the camera owners didn’t play enough with workflow or didn’t experiment enough with REDRAW to feel what is actually possible with REDRAW.

First taking 4K full frame data and scaling it down to 1080p with high quality transforms is GPU/CPU very intensive. To do it right to get all advantages of such downresing and even enhanced dynamic range it takes fast GPU.
Also I had much better results transforming 4K to 2K then 4K to 1080p, don’t want to speculate why but I guess 4/2 gives even numbers.

Second, in broadcast you can’t just shoot 4 minutes or even 15 minutes segments and keep flipping the CF cards conveniently.
Most probably you will use REDDISK anyway. So what the difference if you store 12 hours of 1080p or 2 hours of 4K footage.

Also storage capacity doubles every 6 month or so.

Third, you don’t have to wary about 4K workflow if you need 1080p only.
Sure manipulating data in full 4K is bit better then manipulating data in 1080p but if you do not have system powerful enough just stick to 1080p. And after all if you are asking for 1080p scaled from 4K than you don’t have 4K to begin with.


Now get yourself old $1000 Mac with ATI X1900, activate the script that will take automatically content of REDDRIVE and convert it to 1080p and store the output somewhere on your LAN or on second hard disk.

By the time you finish your lunch your data is ready for rendering on your Mac or PC based post system.
You would like to add CA and other lenses distortion correction while downresing so this process is very usefull out of the camera anyway.
I call it RAW pre-processing.

Mean time if you are in rush and need to see right away what is there use QT proxies.

So no need to have 4K post system if you need just 1080p. Just use, off the camera GPU/CPU system rather than, in the camera one.

Even if we have in the cameara downers capability, right away I see we will need the choice of how we would like to downers the frames, some fast moving/panning scenes needs a different algorithm and some dark scenes need yet the different one.
I hear all these angry voices complaining that downressing sucks and is not working properly all the time.

Also ability to correct for CA and fringing as well for vignetting and wide angle lenses distortion alone makes excellent argument to do all this pre processing out of the camera.

Don’t ask to have 50% of post in the camera, sure build in SCRATCH would be nice.

I see the urgent need for FAQ for post and workflow.

 

[Derek Wiesehahn]

First taking 4K full frame data and scaling it down to 1080p with high quality transforms is GPU/CPU very intensive. To do it right to get all advantages of such downresing and even enhanced dynamic range it takes fast GPU.
Also I had much better results transforming 4K to 2K then 4K to 1080p, don’t want to speculate why but I guess 4/2 gives even numbers.

Second, in broadcast you can’t just shoot 4 minutes or even 15 minutes segments and keep flipping the CF cards conveniently.
Most probably you will use REDDISK anyway. So what the difference if you store 12 hours of 1080p or 2 hours of 4K footage.

Also storage capacity doubles every 6 month or so.

Third, you don’t have to wary about 4K workflow if you need 1080p only.
Sure manipulating data in full 4K is bit better then manipulating data in 1080p but if you do not have system powerful enough just stick to 1080p. And after all if you are asking for 1080p scaled from 4K than you don’t have 4K to begin with.


Now get yourself old $1000 Mac with ATI X1900, activate the script that will take automatically content of REDDRIVE and convert it to 1080p and store the output somewhere on your LAN or on second hard disk.

By the time you finish your lunch your data is ready for rendering on your Mac or PC based post system.
You would like to add CA and other lenses distortion correction while downresing so this process is very usefull out of the camera anyway.
I call it RAW pre-processing.

Mean time if you are in rush and need to see right away what is there use QT proxies.

So no need to have 4K post system if you need just 1080p. Just use, off the camera GPU/CPU system rather than, in the camera one.

Even if we have in the cameara downers capability, right away I see we will need the choice of how we would like to downers the frames, some fast moving/panning scenes needs a different algorithm and some dark scenes need yet the different one.
I hear all these angry voices complaining that downressing sucks and is not working properly all the time.

Also ability to correct for CA and fringing as well for vignetting and wide angle lenses distortion alone makes excellent argument to do all this pre processing out of the camera.

Don’t ask to have 50% of post in the camera, sure build in SCRATCH would be nice.

I see the urgent need for FAQ for post and workflow.

Thanks Andrew, some useful thoughts which raise several questions.
Since I don't have a camera yet, and have no experience with the workflow, can someone tell me about how long it takes to downrez 1 minute of 4k footage to decent quality 1080P HD? Is it necessary to go first to 2k and then to 1080 to achieve the best results? Sounds like something an editor will have a cow over.
What are the current problems with downrezing which people are complaining about?
We could use a post and workflow FAQ here.

 

[Andrew M.]

I think the discrepancy has been the difference between cropped and scaled, and perhaps when people don't indicate which they prefer, wrong assumptions are made. Have you had any requests for a 1080 *cropped* solution?

I definitely appreciate the commitment to continually expanding the feature set.

The only where 1080p cropped solution make sense is for people who need real time 1080p 4:4:4 out via HDSDI link for recording of 1080p.
For monitoring in 1080i/p 4:2:2 I still believe we get it one day on the camere. For monitoring we don’t need sophisticated transforms.

 

[Andrew M.]

Thanks Andrew, some useful thoughts which raise several questions.
Since I don't have a camera yet, and have no experience with the workflow, can someone tell me about how long it takes to downrez 1 minute of 4k footage to decent quality 1080P HD? Is it necessary to go first to 2k and then to 1080 to achieve the best results? Sounds like something an editor will have a cow over.
What are the current problems with downrezing which people are complaining about?
We could use a post and workflow FAQ here.

Last time I have heard, SCRATCH with new Intel MB can do it in ½ the real time.

But most $1000 old used Mac solutions will do 4K to 1080p simple transform at one frame per second.

You have to add the software cost, that could be the one that you use for post anyway.
You don't need to go to 2K first.

What I do right now I take 4096x2304 frame and downscale to 2048x1152 and then I crop it to 1920x1080. Looks perfect. I use Adobe AE.
In AE don’t use scaling, scaling is good for fast monitoring of the 4K output on 2K screen.
Open the new Composition that is 2048x1152 and nest the 4K one in it, AE will use then transform.
You could crop 4096x2304 first and then scale 2:1 but I like to store my output at 2K even if delivery will be at 1080p since who knows maybe I will need it at 2K in the future.

I guess there is better and faster software out there that can do 4K to 1080p without cropping and taking advantage from slight oversampling and uneven pixel shift but I don’t have 50K to spend for extra 2% of better result.

 

[Tony Lorentzen]

We are committed to giving a 1080P (cropped) solution. We are also committed to finding new ways to give our customers more options.

So - no scaled 1080p in a separate firmware build with 4:4:4 1080p out of the dual HD-SDI ports?

 

[Jeff Coatney]

I find it incredible that there's still a desire to have instant 1080p at the expense of all that fantastic over-sampled resolution. The cost of down-converting in post all your day's RED footage is a fraction of what a shooting day costs. It doesn't even come close, it's such a non-issue. I just don't get it. You can't get instant 1080p (or sound for that matter) from a Panavision, Arri, or any other 35mm camera. If you need a video camera, buy or rent a Video camera.

If time is that critical a factor in your business model, multiply your processing capability. Instead of one workstation to do the down convert, add two, three, four machines to the equation on set. The computers needed are the cheapest link in the chain.

But the last thing that should happen is limiting the camera. The camera should always be ahead of the curve. Remember also that RED's resources, however deep, are limited. Time spent moving the camera's capabilities backwards is time away from moving the camera's capabilities forward-- in short, a bad idea.

The Post workflow component should be made to shake the dust off and chase the Camera component and solve this issue with more robust development and faster market response. Jim has set the bar high, but his team reached it. The providers of Post solutions need to step up to the plate now and solve these issues.

Welcome to the 21st Century. The days of using "finished" technology are over. This is the world of "how fast can you run?" Seriously, buy more computers if you need the lower rez footage faster, but don't keep bugging Jim to hamstring his camera.

 

[Andrew M.]

So - no scaled 1080p in a separate firmware build with 4:4:4 1080p out of the dual HD-SDI ports?

I can understand this demand.
HD-SDI out at 1080p 4:4:4 is tempting so you can record to other broadcast like equipment.

You could use multiple cameras for broadcasting.
Instantly RED becomes compatible with thousands of boxes out there like any other broadcast camera.

However after playing with the post I realized that you can get 1080p 4:4:4 out using relatively simple Mac or PC. OK simple with 4:2:2 but put $2K more and you get 4:4:4

Well, not scaled 1080p 4:4:4 out via HD-SDI is possible according to what I red here on this forum, so probably RED will make it as a option so then RED could be used in real time as a broadcast camera in multi camera setup.

 

[Steve Sherrick]

The Post Workflow Event in Vegas will touch on these issues of scaling 4K to 1080P or 2K. I used to think that it would be essential to do this in-camera but now I feel it's better to leave to post due to the various scaling algorithms that can be employed in post to achieve good results on a per shot basis. A one size fits all approach in camera may not achieve the best results. If they could have achieved it though, it can't hurt to have that as an option, so you can choose to do in-camera or in post. But it's a technical limitation.

 

[Brent J. Craig]

Actually, any time someone from Red has replied to 1080 scaled questions, they always went on about how we should all shoot 4K and why would anyone ever want to shoot 1080.

They have NEVER specifically said: NO, this camera cannot now and will never be able to shoot 1080 using the full 35mm sized sensor.

I suspect they are afraid to admit that 1080 scaled is beyond the capabilities of the current camera's processing power. Graeme has made excellent points about aliasing and stuff but if the camera can't do it, please just tell us!

Many of us shoot a lot of HD and would love to shoot HD on a Red using the full chip. I fear that we will all keep asking for it until such time as someone officially says the camera can't do it and will never do it. Not "Why would you shoot 2K when you can shoot 4K". Not "2K cropped lets you use smaller lenses".

If the camera can't do it, it can't do it. We are all grown-ups here. Tell us the truth and we will deal with it. We got over the whole Easter Bunny thing, we will get over this.

Red? 1080 scaled. Yes or No?

 

[Häakon]

They have NEVER specifically said: NO, this camera cannot now and will never be able to shoot 1080 using the full 35mm sized sensor.

That's what Jim has clarified as of yesterday on the first page of this thread.

 

[Ken Willinger]

What Easter Bunny thing??!!??