HDRx can't cope with flames

[Antony M]

Maybe this should have been expected. But it's still a shame as this was the first time I have actually needed HDR from the Epic.

The problem is obviously the motion of the flame differs during the 2 exposures.

A frame:

X frame:

Simple Blend:

Magic Motion:

 

[Matthew Scott]

Well, magic motion can't cope with it. You could cope with it in DaVinci, or any software that allows you access to both tracks using secondaries to select the parts you want. Cool shot BTW

 

[Andrew Allsbury]

Would you mind sharing this r3d still? I would like to try something out. Thanks!

 

[Antony M]

Well, magic motion can't cope with it. You could cope with it in DaVinci, or any software that allows you access to both tracks using secondaries to select the parts you want. Cool shot BTW

You are obviously better than me
I tried (and failed) at combining them in After Effects and Photoshop (for stills).

The shots were a lighting test for a music video we are shooting in a month or two.

 

[Antony M]

Would you mind sharing this r3d still? I would like to try something out. Thanks!

Of course.

It should be at http://www.sendspace.com/file/qncwmu
Let me know if the link doesn't work.

 

[Antony M]

OK, well I got to this with a layer mask and curves in Photoshop, but not really a practical workflow for a video clip:

 

[Andrew Allsbury]

My take.

 

[John Schick]

I tried blending the images in After Effects using linear light:

It's not perfect, but it might work. The only way to be sure is to see it in motion.

I tried finding a balance between the A and X track. I think the exposure difference might be too extreme - you could try shooting it again with a lower HDRx value.

 

[Elsie N]

Are there any filters, coupled with using HDRx, that can capture fire?

 

[Björn Benckert]

It may be so that when you see any of the above in motion it looks good. And you should try to have less stops on the HDRx (the less the better interpolation).

 

[Karim D. Ghantous]

Antony, it looks like your simple blend is exactly what I got with the GIMP. I just had each frame as a separate layer and altered opacity. This one is at 65% opacity:

Note that it doesn't have much saturation. Anyway you can writ scripts with the GIMP. The problem is that I don't know how to, and plus, the image would need something else. How about HDR software? It would have to be scriptable though. Or at least, have batch processing capability.

 

[Naim Sutherland]

Try this: http://vimeo.com/24636714

 

[Antony M]

Try this: http://vimeo.com/24636714

Some great tips in that video.

But they don't work in this case, because when you key out the overbright (whether using the Luma matte in the video or by any other means), due to the motion difference, there is nothing in the other layer to take its place.

In his beach examples, there is always a beach on both A and X. In my footage, there isn't always a flame on X, so you get odd black patches:

 

[Antony M]

My take.

That's very good.
Still a bit of "ghosting" in parts, but would probably not be noticed in motion.
Did this involve much manual masking, or could it be applied to the whole video clip?

 

[Peter Majtan]

You really need to be grading this in motion. A lot of artifacts seems extreme when You look at still and are but invisible in motion - and vice versa...
I also second to use max of +3 HDRx mode....

There is no time difference per say in HDRx. Think of it as if they both started at the exact same time. The difference is that one (the X-track) ends much sooner. How much sooner is determined by the HDRx factor. The larger the HDRx factor - the larger the difference between the ends of the exposure. So if You have something deforming organically as much and as fast as a burning flame - You will start notice the differences. Having said that - in such a organically dynamic imagery as a burning flames - these "strange" gaps are actually something rather natural - this is how flame behaves. I am certain that it will not look weird to you once you see it in motion...

Peter

 

[Marc Wielage]

Fire is always hard. My general advice is to do tests beforehand, and provide lots of fill (even at night), then drop the blacks in the final correction and expose for the highlights. Once the highlights are sheared off, there's no detail there. Underexposing about 1/3 of a stop is generally a good idea. That plus fill will avoid crushing the exposed image.

We did that for years on Lost, and it worked very well. I never had a show with so many campfire/torchlit scenes in my life, and it was challenging to make that work. The results looked great, and that was all an ISO 500 show (Kodak 5219 film).

 

[Karim D. Ghantous]

The results looked great, and that was all an ISO 500 show (Kodak 5219 film).

This would explain why I read (somewhere) that film is generally preferred for explosions. I wonder how you would have coped if you were forced to use video cameras?

 

[paulherrin]

give me the r3d and i'll give you a shot you can use.

 

[Pablo Moreno]

You really need to be grading this in motion. A lot of artifacts seems extreme when You look at still and are but invisible in motion - and vice versa...
I also second to use max of +3 HDRx mode....

There is no time difference per say in HDRx. Think of it as if they both started at the exact same time. The difference is that one (the X-track) ends much sooner. How much sooner is determined by the HDRx factor. The larger the HDRx factor - the larger the difference between the ends of the exposure. So if You have something deforming organically as much and as fast as a burning flame - You will start notice the differences. Having said that - in such a organically dynamic imagery as a burning flames - these "strange" gaps are actually something rather natural - this is how flame behaves. I am certain that it will not look weird to you once you see it in motion...

Peter

Can you explain how it works?, is HDRx 48fps alternate, or it works some other way??

 

[Antony M]

Can you explain how it works?, is HDRx 48fps alternate, or it works some other way??

Essentially it works by reading the sensor twice.

For example, if you are shooting a 2-stop HDR at 25fps with a 1/50 second exposure, then a reading is also taken after 1/200 second.