More HDRx...

[joshua csehak]

That's incorrect. The HDRx™ exposure is a subsection of the main exposure and it's at the beginning not the end. As to whether the normal exposure includes the HDRx™ timeframe or the two are split and then averaged back together to form the "normal" exposure even without Magic Motion is unknown and really ultimately irrelevant since they would look exactly the same if it was in fact possible to do that without a gap. I doubt it's possible without a gap. That's why I'm betting the trick they've pioneered is probably the early readout without interrupting the normal exposure and it sounds like that's a cool hardware trick they had to build into the chips to achieve.

Jim also said it's only possible to have the HDRx™ read at the beginning not the end. So that also discourages the idea that they are two separate frames stitched together and not an early read of the same exposure. If it was just recording another frame without a gap they could do it arbitrarily anywhere in the frame (even centered). The fact that it's only the very first part is a strong hint that it's two reads of one exposure at different times through the exposure.

Now I'm confused. If you look at the spin clip, she's spinning the pole clockwise (relative to the viewer), so the shorter exposure is clearly taken at the tail end. Unless I'm missing something? But it really does look like it's taken at the same time as the regular exposure, cause they seem to completely overlap. Which is pretty amazing.

 

[Wolfgang Woehl]

Joshua, where's the pole at the beginning of the exposure ...

 

[joshua csehak]

Joshua, where's the pole at the beginning of the exposure ...

Haha, I'm an idiot. Or maybe just dyslexic. Thanks, Wolfgang.

I guess I remembered the article wrong. I was looking for it to check, but gave up after looking for 5 minutes.

 

[I Bloom]

The faster underexposed stream has no motion blur on it. So unless you want your sky razor sharp and your mountains motion blurred you have to blur the sky to match.

In order to blur the sky you have to figure out which direction to blur the sky in. That's where oFlow comes in. It tracks the sky. Determines what direction and how fast the sky is moving and then applies motion blur to the sky pixels.

In that same case if your sky is all white you only have one image to track: the HDRx™ underexposed image. So the Foundry can't assume that there is any useful data in the normal exposure. Not only is the motion blurred normal exposure less ideal for optical flow, it's also potentially clipped to white which would be as far as the tracker is concerned unmoving.

In the exposure ranges where both images are visible the optical flow algorithms could theoretically track both to get a better optical flow track than either single image sequence but if you already are within the normal exposure range of both images there is no reason to use the HDRx™ image: Just use the good exposure. The HDRx™ is most useful in the parts of the normal image which are overexposed. The parts of the normal image which are overexposed are useless for tracking so the tracking will almost certainly be almost entirely if not entirely using the HDRx™ image (which is a better source for the track anyway).

And again. Even if the optical flow track were somehow perfect (which nobody has ever done nor probably ever will do for a long long time) the limitations on post production blurring wouldn't go away.

There are theoretically some interesting/cool things that The Foundry could do with both a motion blurred and a normal stream when they can compare pixels at the same point. But the whole point of HDRx™ is to extend the dynamic range beyond one of the image streams. So everywhere that HDRx™ is useful is the same place where you only have one stream of data to analyze.

I think you are making many many assumptions here that probably are not true. I don't think that RED is using optical flow to do their basic motion adjustment, so most of your arguments seem irrelevant to me. The occlusion problem that you talked about with the puppy and the truck also doesn't apply because the highlight track is topologically "contained" by the blown out portions of the lowlight track. Whenever it isn't we have information in the lowlight track and don't need the highlight track. The puppy would not be blurred because we have its mask essentially. Again this is all because the highlight track is recorded at the beggining of the exposure- a very special case.

IBloom

 

[Bill Anderson]

There had been a few predictions based upon a premise that data, in order to
get from A to Z, must ( or perhaps does) pass through other states, and that it might be effective to trap one of those early states, i.e. "A", in order to apply it to the final state "Z". Could that be qualified as "two separate" exposures? I think so.
But, like I said earlier, since I have more imagination than knowledge in these affairs, how accurate is this? And is it possible at the rate at which light and sensor data rates interact? Easier said than done.

 

[Gavin Greenwalt]

I think you are making many many assumptions here that probably are not true. I don't think that RED is using optical flow to do their basic motion adjustment, so most of your arguments seem irrelevant to me. The occlusion problem that you talked about with the puppy and the truck also doesn't apply because the highlight track is topologically "contained" by the blown out portions of the lowlight track. Whenever it isn't we have information in the lowlight track and don't need the highlight track. The puppy would not be blurred because we have its mask essentially. Again this is all because the highlight track is recorded at the beggining of the exposure- a very special case.

IBloom

No I'm not saying they're using optical flow for Magic Motion. Just for the MNMB effect.

In the puppy scenario I'm referring to a scenario where the puppy/truck pixels are clipping to white in the normal exposure. In which case you only have the blur free HDRx channel for data.

A white delivery truck driving through a window like this would go white in the normal exposure:
http://www.fcccambodia.com/newsletter/0809/media/reflections-500.jpg

But that's a more extreme example that you'll encounter less often. Much more frequent would be a scenario where you have someone inside a car or train. You're going to lose their reflection since it'll be blurred out completely with MNMB. Everything blown out in the window will pick up the motion blur from the exterior. The reflection will get motion blurred along with the passing scenery.

 

[Álex Montoya]

No I'm not saying they're using optical flow for Magic Motion. Just for the MNMB effect.

In the puppy scenario I'm referring to a scenario where the puppy/truck pixels are clipping to white in the normal exposure. In which case you only have the blur free HDRx™ channel for data.

A white delivery truck driving through a window like this would go white in the normal exposure:
http://www.fcccambodia.com/newsletter/0809/media/reflections-500.jpg

But that's a more extreme example that you'll encounter less often. Much more frequent would be a scenario where you have someone inside a car or train. You're going to lose their reflection since it'll be blurred out completely with MNMB. Everything blown out in the window will pick up the motion blur from the exterior. The reflection will get motion blurred along with the passing scenery.

That's cool, because I find MagicMotion very pleasing and, knowing that it may well be artifact-free, I will probably use it more often than not.

 

[Gavin Greenwalt]

That's cool, because I find MagicMotion very pleasing and, knowing that it may well be artifact-free, I will probably use it more often than not.

If you like the look of MagicMotion I'm assuming it will not have any optical flow (A very safe assumption since they can do it in camera). (But) MagicMotion won't be great for shots with large blown out areas such as car windows since the blown out areas will all be razor sharp and motion blur free. So you won't get a MagicMotion mix between blur and sharp you'll just get sharp (the only channel which has any information in those regions). Although if you put some NDs on shot 360* shutter for the normal interior exposure and then only used +2 HDRx™™ exposure the exterior would be HDR at 90* shutter which is still a bit blurry. So it could be done but you would need to adjust your shooting a little bit and you would have a bit smeary interior (which would be pretty static so that would probably be ok).

 

[Álex Montoya]

Good points, Gavin. Of course, you have to get to know your tools.

I'm just saying it's a bit early to get too picky.

 

[Joe G.]

"No I'm comparing it to a currently non-existent 'normal' camera which could capture 18 stops natively."

No, you're not. You're not camparing "it" since you don't have it to compare and it's not done yet. You also have no idea what's under the hood, but are merely speculating.

You could at least wait until it is done before you pronounce its repeated failure.

 

[Jannard]

Everyone who has seen it is surprised at what it does and surprisingly what it doesn't do, based on their preconceived notions. I suspect that will continue to be the case.

Don't forget that there are many layers of options here. You can, depending on your skill or budget, use masks and do some pretty involved stuff with the "x" track, including motion and grading. You can adjust the amount of the "x" track you use in REDCINE-X. You will even have some limited control options in camera with eHDR. It involves knowledge (not speculation) of what each does. That knowledge is what separates a "pro" from a "prosumer". Testing will be necessary, and I have found out... quite interesting and fun. Pushing along in new territory is something that keeps us moving forward and not getting stale.

One thing I agree with is that nothing is perfect. But eHDR and HDRx sure solve a lot of problems and give a ton of creative choices and capabilities that we have never seen before. I like it.

Jim

 

[David Baumber]

Jim, You're very active on the forums on Sunday's. Do I take it you're sitting in a robe with the morning paper and smoking a fat cigar before popping down to your local church? I'm praying that you send us some new images today and I'm not a religious man!

 

[I Bloom]

If you like the look of MagicMotion I'm assuming it will not have any optical flow (A very safe assumption since they can do it in camera). (But) MagicMotion won't be great for shots with large blown out areas such as car windows since the blown out areas will all be razor sharp and motion blur free. So you won't get a MagicMotion mix between blur and sharp you'll just get sharp (the only channel which has any information in those regions). Although if you put some NDs on shot 360* shutter for the normal interior exposure and then only used +2 HDRx™™™™™ exposure the exterior would be HDR at 90* shutter which is still a bit blurry. So it could be done but you would need to adjust your shooting a little bit and you would have a bit smeary interior (which would be pretty static so that would probably be ok).

Yup, sorry Gavin. I think all of the information is actually there. You are thinking very conceptually as a compositing problem when this is in fact a pixel level image processing problem.

If the car window was completey blow out then yes you are right. Not saying there aren't situations where there will be artifacts. Just that it's possible to do it well with less "intellegent" algorithms and very consistent results. I'm not selling my Rosco stock yet though.

Hopefully we can see the result while we are shooting so that we might adjust?

IBloom

 

[Shawn Nelson]

I wonder if using HDRx during the day could help in giving a more "realistic" day for night effect?

 

[Martin Stevens]

Please post side-by-side comparisons with 35mm Negative movie FILM SCANS when you have the time.
No rush, just something I think is vital to convince everyone that not even film has 18 stops of DR.......ah, or does it?

 

[David Quinn Carder]

That knowledge is what separates a "pro" from a "prosumer". Testing will be necessary, and I have found out... quite interesting and fun. Pushing along in new territory is something that keeps us moving forward and not getting stale.

Yes! That's what I love about RED. Beyond simply matching film, it is letting us do new things. I really can't wait to see some of the as-of-yet not-all-that-talked-about possibilities on screen.

I wonder if using HDRx™ during the day could help in giving a more "realistic" day for night effect?

Isn't one of the advantages of RED's awesome low-light performance that it gives you a more realistic night for night effect? :smiley:

 

[Nikhil Kamkolkar]

... It involves knowledge (not speculation) of what each does. That knowledge is what separates a "pro" from a "prosumer". Testing will be necessary, and I have found out... quite interesting and fun. Pushing along in new territory is something that keeps us moving forward and not getting stale.
Jim

And as an indie filmmaker, that pretty much sums up what RED is about to me. Going way beyond that tired old idea of "recreating film digitally", and into the unknown territory of brand new imagery that has nothing at all to do with film. Those who wish to get the film look should and will just shoot film... RED is beyond film.

 

[Jason Sinclair]

I suppose the big question on the lips of those who are planning options for the future in terms of getting solutions for dynamic range increase is will there be consistency? I am not talking about other solutions as in cameras. It is obvious nothing comes close to the Epic.

Using a setting and having the choice is great but it begs the question of matching shots if there are cases where it cannot be employed, especially in low budget features. Having little knowledge of the processes employed certainly, one cannot make conclusions and will have to wait and make the suitable tests.

My feeling is that if the hardware is set and the future changes involve software and firmware upgrades then it is advisable to go forward with Epic production even in the case of HDRx™ isn't fully complete.

We all love the firmware upgrades in red and the evolving processes. It is not like this will be a problem to anyone. Hadrware upgrades, well that is a different story.

HDR is important from a film makers perspective and the promises it may hold under strenuous conditions very valuable but I think it is folly thinking that this camera without HDRx™ is still not revolutionary and will change the industry as we know it.

Either way, if we wait for HDRx™ to be finished first or vice versa we still will be on the same schedule only have Epics in the field sooner rather than later. That is good for us and also good for Red as it starves any incoming competition.

 

[David Mullen ASC]

Keep in mind that wider dynamic range can actually lead to more visual inconsistency, simply because you are able to get an image in a wider variety of situations, and filmmakers being who they are, will take advantage of that.

I noticed this to a smaller degree when faster film stocks came out that were also lower in contrast -- you saw more visual inconsistency because the color and level of ambient fill that reality provided varied shot-by-shot, and this difference was showing up, whereas on the older, slower, more contrasty stocks, you basically had to light for what you wanted to see in the shadows.

In other words, the need to know how to control the image doesn't go away just because you have better dynamic range -- all that gives you is more options and more flexibility in shooting, more creative control... but at the expense of more things to keep track of, more variables to choose from, and a greater likelihood of having to adjust the image in post to get things to match shot to shot, since one of the downsides of capturing more natural light is that you discover just how much it varies minute by minute throughout the day. For example, one moment a room may be illuminated by the sun bouncing off of a yellow house across the street and yet an hour later, the room may mostly be lit by blue-ish skylight, a radical change of color and intensity.

 

[Jannard]

Keep in mind that wider dynamic range can actually lead to more visual inconsistency, simply because you are able to get an image in a wider variety of situations, and filmmakers being who they are, will take advantage of that.

I noticed this to a smaller degree when faster film stocks came out that were also lower in contrast -- you saw more visual inconsistency because the color and level of ambient fill that reality provided varied shot-by-shot, and this difference was showing up, whereas on the older, slower, more contrasty stocks, you basically had to light for what you wanted to see in the shadows.

In other words, the need to know how to control the image doesn't go away just because you have better dynamic range -- all that gives you is more options and more flexibility in shooting, more creative control... but at the expense of more things to keep track of, more variables to choose from, and a greater likelihood of having to adjust the image in post to get things to match shot to shot, since one of the downsides of capturing more natural light is that you discover just how much it varies minute by minute throughout the day. For example, one moment a room may be illuminated by the sun bouncing off of a yellow house across the street and yet an hour later, the room may mostly be lit by blue-ish skylight, a radical change of color and intensity.

David... agree with you (again). I think half the battle will be deciding when to use HDRx™ and when not to. The variable in the equation is that EPIC will have even less noise and FPN (when pushed to the extreme) than you have ever seen.

It is great to have lots of tools. This time, they will all be sitting inside the same camera and you get to decide which and when you will choose what to use.