More HDRx...

[Bill Anderson]

And it tells you even more about the "analyzer".

 

[Karel Bata]

And it tells you even more about the "analyzer".

And that says so much about you.

 

[Sanjin Jukic]

I doubt the xTrack actually contains a vector pass. Not sure where that data would be coming from. I would assume the xTrack is just a short exposure. The foundry can extract vectors from that, and by being a normal image you get more flexibility in the future when algorithms improve. Baking in vectors would lock you in, which seems pretty un-red-like.

But yes Sanjin you are wasting your time by splitting up your image into two and then recombining it. Just apply local tone mapping. That's different from curves that you tried to use as an example of being unable to reproduce "HDR". Curves or LUTs are global tone mapping. Local tone-mapping adjusts contrast based on spatial filters in addition to luminance and chroma. An example of a local tonemapping algorithm would be the "Shadows/Highlights" tool in Photoshop.

Behold the power of tonemapping on a single exposure from REDCine. Same nice and overprocessed look, half the work.:

Gavin very nice comment but I do have doubts but I can reply on that later because busy at the moment.

But just a short comparison between our two examples:

Gavin's example (onemapping on a single exposure from REDCine).

Sanjin's example made from two different exposures HDR generated in Photomatisx Pro.

Which image has more DR?

 

[Jeff Kilgroe]

Which image has more DR?

DR is a property of the acquired image, not the processed image. They are both the same, it's a matter of processing. You can tone-map and alter luminance / exposure any way you like. Even on a per-pixel, or sub-pixel level. That does not extend DR, it only molds the data you have to work with initially.

 

[Michael Olsen]

I really don't want to denigrate what others are doing...

But is it possible to move the discussions of tone-mapping images after acquisition via multiple files or highlight/shadows and so forth into another discussion?

Don't get me wrong - it is very interesting and I have read it all, but it seems to detract from the point of this thread: HDRx.

And I think we can all agree - HDRx is not the same as creating multiple images in post and recombining them or pushing the shadows and highlights of footage that was already shot, as useful and representative of the M-X's dynamic range as those techniques may be.

 

[Elsie N]

I really don't want to denigrate what others are doing...

But is it possible to move the discussions of tone-mapping images after acquisition via multiple files or highlight/shadows and so forth into another discussion?

Don't get me wrong - it is very interesting and I have read it all, but it seems to detract from the point of this thread: HDRx™.

And I think we can all agree - HDRx™ is not the same as creating multiple images in post and recombining them or pushing the shadows and highlights of footage that was already shot, as useful and representative of the M-X's dynamic range as those techniques may be.

I wonder if all the relevant points to HDRx weren't covered about 30 pages ago?

 

[Gavin Greenwalt]

I wonder if all the relevant points to HDRx™™™™ weren't covered about 30 pages ago?

Try and offer an opinion on what HDRx™™™ is and you are accused of trying to give away RED's secrets to their competitors.

Try and speculate on when it would and would not work and you're accused of trashing the idea as a "complete failure".

For some there is no acceptable posting except 30 pages of "Wow! Perfect! I just wet myself! I'm not interested in understanding what I'm seeing so please don't explain anything!"

Not that I'm disagreeing with Michael. It is pretty off topic but we seem to have run out of topics which are on topic that can be discussed without eliciting condemnation as being some sort of RED Judas.

 

[Sanjin Jukic]

Also I do not get why some people getting dizzy here on this tread where we are discussing HDR

and also HDRx but we can't say anything more from this side about that at all.

Normally that RED wouldn't tell us more about details of HDRx anyway.

Some people try to play a sort of HDR experts but they are really not.

Only Graeme Nattress is an expert for HDRx but unfortunately he must stay mostly silent about that.

 

[Elsie N]

Try and offer an opinion on what HDRx™™™™ is and you are accused of trying to give away RED's secrets to their competitors.

Try and speculate on when it would and would not work and you're accused of trashing the idea as a "complete failure".

For some there is no acceptable posting except 30 pages of "Wow! Perfect! I just wet myself! I'm not interested in understanding what I'm seeing so please don't explain anything!"

Not that I'm disagreeing with Michael. It is pretty off topic but we seem to have run out of topics which are on topic that can be discussed without eliciting condemnation as being some sort of RED Judas.

Gavin, have read your posts with interest. I just saw a curious mind racing frenetically trying to figure things out. That said, I do think you are looking at it from a perspective that probably affects your line of work mostly. And based on explanations, it seems your concerns over the direction of motion blur etc. maybe are being addressed by the Foundry if not by Graeme himself. Just wondering if you are not a little premature in your critique? Not saying your points are not valid. Just saying that only the final version will determine that.

Still, as I said... I for one read your posts. But even if you are right that in some number of instances things are not perfect, as long as we know any such limitations we can operate at otherwise peak efficiency as users. And I think you made a similar statement way back on page number ?.

 

[Hrvoje Simic]

:jacked:
:beatdeadhorse5:

:emote_popcorn:

 

[Gavin Greenwalt]

Still, as I said... I for one read your posts. But even if you are right that in some number of instances things are not perfect, as long as we know any such limitations we can operate at otherwise peak efficiency as users. And I think you made a similar statement way back on page number ?.

Haha I think I said that on page 2. All I said originally was that a real HDR sensor was still be useful since there would be use cases where HDRx™™ wouldn't work.... and the rest is history.

As to being premature. It's not really premature because while this is the first time an optical camera has done this to my knowledge the tech behind it on the post side and the technique are common to CG and VFX.

There are also just fundamental image processing principles at work that apply to all techniques. It's almost impossible for instance to synthesize missing data (blurring out a foreground element and revealing what was behind it). The CSI "enhance" joke where they remove a foreground element to see what was behind it. Or it's also currently mostly impossible to separate semi-transparent images without modifying both layers except in very specific controlled situations. They're like the "conservation of energy" laws for image processing. No matter what technique you use you have to obey the laws of physics.

There are reasons why VFX artists prefer 3D motion blur to 2D motion blur because of all the glitches that arise in a perfect scenario where you have subpixel samples of color, depth, velocity and even background data. If it hiccups in spots with more data than a camera could ever hope to capture it's not much of a stretch to say that an imperfect, real world environment is going to be even more troublesome.

Others aren't a stretch because in the case of a blown out window you know the only data at your disposal is in the xChannel in which case it's not really anything new in that region, it should be exactly the same as shooting a fast shutter video right now since you're losing the 'x' unique advantage of two channels to reference.

You don't really have to wait. Downloads Reel Smart Motion Blur or load up furnace F_MotionBlur and play with some fast shutter footage. It'll give you an idea of what it's like and when it'll work and when it won't.

MagicMotion though is another beast, but Magic Motion is only Magic Motion in the regions which don't clip. After that you just have 'x'. So again you can see what large clipping areas would look like with magic motion, just take a fast shutter clip. If you so overexposed an HDRx™™ shot so much that it was completely clipping in the normal channel all you would have is HDRx™™ and it would look like a fast shutter video (as far as we know) today. So with a little imagination you can picture what a portion of the frame would look like (fast shutter shot).

HDRx™™ is special and unique because you have two tracks of video that are temporally aligned. But neither track by itself is anything magical as far as I can tell and still subject to the same laws of image as any other footage.

But if nothing is moving very quickly then it should be for all intents and purposes near perfect HDR. And thankfully most films are composed of people standing relatively still and talking. And considering most people have managed to expose films properly without HDRx™™ all these years I suspect the % of frame being HDRx™™ is going to be minimal in which case MagicMotion would work in almost every situation. For shots that already worked HDRx™™ is going to probably work great. I just raise an eyebrow when people talk about things like shooting out of a car window. The more impossible the exposure used to be the more restrained you're going to need to be in your movement.

(All assuming of course the x track is the first few milliseconds of the exposure read off early which it pretty distinctly would appear to be in all shots posted).

 

[Roberto Lequeux]

Nice grade on that one SJ!

Sorry if I am contributing to derailment, I just really like that frame. And I agree, this thread is a goner so it can't hurt much.

 

[Elsie N]

Haha I think I said that on page 2. All I said originally was that a real HDR sensor was still be useful since there would be use cases where HDRx™™™ wouldn't work.... and the rest is history.

As to being premature. It's not really premature because while this is the first time an optical camera has done this to my knowledge the tech behind it on the post side and the technique are common to CG and VFX.

There are also just fundamental image processing principles at work that apply to all techniques. It's almost impossible for instance to synthesize missing data (blurring out a foreground element and revealing what was behind it). The CSI "enhance" joke where they remove a foreground element to see what was behind it. Or it's also currently mostly impossible to separate semi-transparent images without modifying both layers except in very specific controlled situations. They're like the "conservation of energy" laws for image processing. No matter what technique you use you have to obey the laws of physics.

There are reasons why VFX artists prefer 3D motion blur to 2D motion blur because of all the glitches that arise in a perfect scenario where you have subpixel samples of color, depth, velocity and even background data. If it hiccups in spots with more data than a camera could ever hope to capture it's not much of a stretch to say that an imperfect, real world environment is going to be even more troublesome.

Others aren't a stretch because in the case of a blown out window you know the only data at your disposal is in the xChannel in which case it's not really anything new in that region, it should be exactly the same as shooting a fast shutter video right now since you're losing the 'x' unique advantage of two channels to reference.

You don't really have to wait. Downloads Reel Smart Motion Blur or load up furnace F_MotionBlur and play with some fast shutter footage. It'll give you an idea of what it's like and when it'll work and when it won't.

MagicMotion though is another beast, but Magic Motion is only Magic Motion in the regions which don't clip. After that you just have 'x'. So again you can see what large clipping areas would look like with magic motion, just take a fast shutter clip. If you so overexposed an HDRx™™™ shot so much that it was completely clipping in the normal channel all you would have is HDRx™™™ and it would look like a fast shutter video (as far as we know) today. So with a little imagination you can picture what a portion of the frame would look like (fast shutter shot).

HDRx™™™ is special and unique because you have two tracks of video that are temporally aligned. But neither track by itself is anything magical as far as I can tell and still subject to the same laws of image as any other footage.

But if nothing is moving very quickly then it should be for all intents and purposes near perfect HDR. And thankfully most films are composed of people standing relatively still and talking. And considering most people have managed to expose films properly without HDRx™™™ all these years I suspect the % of frame being HDRx™™™ is going to be minimal in which case MagicMotion would work in almost every situation. For shots that already worked HDRx™™™ is going to probably work great. I just raise an eyebrow when people talk about things like shooting out of a car window. The more impossible the exposure used to be the more restrained you're going to need to be in your movement.

(All assuming of course the x track is the first few milliseconds of the exposure read off early which it pretty distinctly would appear to be in all shots posted).

Before I agree with you (meaning I'm not knowledgeable enough to point out any possible flaws) let's wait until Graeme's hands are untied and he is able to respond. Obviously there are secrets and the only way to preserve them is to keep silent.

I put the barn scene on my desktop as background. I am startled every time I turn on my computer and that image is before me.

In theory everything you posit may be true and correct. In practice, I am so happpeeeee!

 

[I Bloom]

Haha I think I said that on page 2. All I said originally was that a real HDR sensor was still be useful since there would be use cases where HDRx™™™ wouldn't work.... and the rest is history.

As to being premature. It's not really premature because while this is the first time an optical camera has done this to my knowledge the tech behind it on the post side and the technique are common to CG and VFX.

There are also just fundamental image processing principles at work that apply to all techniques. It's almost impossible for instance to synthesize missing data (blurring out a foreground element and revealing what was behind it). The CSI "enhance" joke where they remove a foreground element to see what was behind it. Or it's also currently mostly impossible to separate semi-transparent images without modifying both layers except in very specific controlled situations. They're like the "conservation of energy" laws for image processing. No matter what technique you use you have to obey the laws of physics.

There are reasons why VFX artists prefer 3D motion blur to 2D motion blur because of all the glitches that arise in a perfect scenario where you have subpixel samples of color, depth, velocity and even background data. If it hiccups in spots with more data than a camera could ever hope to capture it's not much of a stretch to say that an imperfect, real world environment is going to be even more troublesome.

Others aren't a stretch because in the case of a blown out window you know the only data at your disposal is in the xChannel in which case it's not really anything new in that region, it should be exactly the same as shooting a fast shutter video right now since you're losing the 'x' unique advantage of two channels to reference.

You don't really have to wait. Downloads Reel Smart Motion Blur or load up furnace F_MotionBlur and play with some fast shutter footage. It'll give you an idea of what it's like and when it'll work and when it won't.

MagicMotion though is another beast, but Magic Motion is only Magic Motion in the regions which don't clip. After that you just have 'x'. So again you can see what large clipping areas would look like with magic motion, just take a fast shutter clip. If you so overexposed an HDRx™™™ shot so much that it was completely clipping in the normal channel all you would have is HDRx™™™ and it would look like a fast shutter video (as far as we know) today. So with a little imagination you can picture what a portion of the frame would look like (fast shutter shot).

HDRx™™™ is special and unique because you have two tracks of video that are temporally aligned. But neither track by itself is anything magical as far as I can tell and still subject to the same laws of image as any other footage.

But if nothing is moving very quickly then it should be for all intents and purposes near perfect HDR. And thankfully most films are composed of people standing relatively still and talking. And considering most people have managed to expose films properly without HDRx™™™ all these years I suspect the % of frame being HDRx™™™ is going to be minimal in which case MagicMotion would work in almost every situation. For shots that already worked HDRx™™™ is going to probably work great. I just raise an eyebrow when people talk about things like shooting out of a car window. The more impossible the exposure used to be the more restrained you're going to need to be in your movement.

(All assuming of course the x track is the first few milliseconds of the exposure read off early which it pretty distinctly would appear to be in all shots posted).

Whatever, I can't take any of your posts seriously anymore Gavin because you keep leaving out one important thing man... the possibility that real magic is being used to fix the motion blur in the x-channel, or small Canadian elves repaint each frame by hand. You can call me a fanboy or whatever... I still want that camera.

IBloom

 

[Jeff Kilgroe]

Come on, Ian. You and I both know Graeme has found a way to cross-breed his magic Canadian pixies with one of the more rare English varieties. I think it's all in the whiskey...

 

[Gavin Greenwalt]

Whatever, I can't take any of your posts seriously anymore Gavin because you keep leaving out one important thing man... the possibility that real magic is being used to fix the motion blur in the x-channel, or small Canadian elves repaint each frame by hand. You can call me a fanboy or whatever... I still want that camera.

IBloom

Damn pixies always ruining my best laid plans! (and stealing my socks).

 

[Eric Haase]

I think theorizing that whatever is outside a car window may only be captured on the x track and thus have a skinny shutter look is an interesting idea. It will be interesting to see of this is the case and makes one think about possible uses and possible trouble areas for hdrx as we currently understand it.

Also wanted to note, as I inferred from the interview on redcenter that perhaps you cannot see the results of hdrx or easy hdr in the viewfinder, but only after the footage has actually been recorded. Not sure if this has been discussed or clarified, but it would be interesting if it were the case.

 

[Michael Olsen]

Not sure if this has been discussed or clarified, but it would be interesting if it were the case.

I don't think it has...questions about monitoring HDRx and eHDR seem to go unanswered. At least for the time being.

 

[Michael Hastings]

Come on, Ian. You and I both know Graeme has found a way to cross-breed his magic Canadian pixies with one of the more rare English varieties. I think it's all in the whiskey...

Is this pixie revelation new or do I have to read through all 600 posts to catch up? Man this new RED technology happens so quick it's hard to keep up!

And what's a "whiskey"?

 

[Gavin Greenwalt]

And what's a "whiskey"?

A racial slur for white pixies. Pronounced "Why-skee"