Produce True HDR video from any camera - NEX5N sample

H

HHL

Veteran
I've been kicking this idea around for 2 years and last night I saw the final piece of the puzzle.
This was a rushed first attempt and considering I've never tried the workflow before, this looks pretty promising.


HDR video on the NEX5N or any other camera on the planet.

(Please note that although this was done using one camera, we're all smart enough to know that to be very useful on a typical shot, it will require 2 cameras and a beam splitter.....we'll get to that after we're sure that the workflow will work satisfactory for all of us pixel peeping nut jobs, lol!)


I just used exposure compensation +3/-3 and some (simple) workflow magic.


Take a look here at the first attempt:




So....here's something that I just want to throw out to the community:

If you'd like to see what YOUR camera would look like if you had an extra 2-4(?) stops of dynamic range...send me 4 seconds of +2 and 4 seconds of -2, send me the link to your upload/file share and I'll process it for you and we'll post it and see what it looks like. How does that sound?


If you don't have time or interest, it's totally cool. I know we're all busy.

Some notes when shooting the footy, (I pushed mine to +3/-3, but I'm not sure if we can get away with it or not)
(shot from a tripod, a "non-moving" subject matter, change the exposure, shoot again)

I have no idea how many folks if any are going to respond, so if I don't have time to process them all, I'll change this first post.

I don't know about you all, but I'm REALLY curious what the Hacked GH2 with Driftwood's latest might look like with an extra 2 stops of DR. :) Maybe an FS100 with the same range as the F3? (That might be tasty) Anywho...We'll see how much the footy holds up from the other cams and we'll go from there. Mad love to DVXUsers everywhere!
 
I've done this before on locked down shots with highlights out of my control. I'll record the scene, then record 10-15 seconds stopped down a little and replace only the blown highlights in the original clip.

However... it's not very useful in most situations unfortunatlely. And since you can't sync the shutters of DSLR cameras, even a beam splitter is probably going to produce iffy results.
 
Andrew McCarrick said:
Don't know if you noticed, but that footage flickers because the shutters are out of sync.
Click to expand...

And that's still a fairly locked down shot. Add any type of serious movement (handheld, jib, dolly, steadicam) and the flicker is going to get worse. Hopefully someday we'll be able to lock two DSLR's together so their shutters are in sync, but until then, this technique is going to be extremely hit or miss.
 
Looks awesome man, keep experimenting! What happens if someone walks through the shot though? Do you have to roto around them?
 
In the long, I really don't see a use for this as a regular tool. It's great if you're going for the hyper-real look for some shots. But HDR to retain highlights and grade it back to a normal look; why not just go for a better camera? A beam splitter plus 2 DSLRs will likely be 8 to 12 grand; that's borderline F3, C300, Scarlet territory.
 
Andrew McCarrick said:
Don't know if you noticed, but that footage flickers because the shutters are out of sync.
Click to expand...

Good observation. I will (in this case) that the flicker is the result of how the NEX5N implements exposure compensation. I notice as I changed the exposure compensation that the camera changed the shutter speed values. The highlight optimized shot had a shutter somewhere around 320, where the low light optimized was around 50 (180 degree shutter). You can tell if you look at the first shot that it has the "proper" motion as you would expect. Look at the other (before they are combined and you'll see the same stutter due to high shutter speed.) The reason I didn't change it is because I thought it would be simpler to express it in written form. (+3/-3) I'll make sure to use ISO next time (also very easy in written form ISO200-3200.) I didn't want to open that can of worms though because people would correctly question the impacts of using high ISO with low ISO and the impacts on the overall image quality. That said, I see that opened the shutter can of worms instead. (Such is life, lol)

I'll do a split screen this weekend of the same shots just to demonstrate. Thanks for the observation.

awelgraven said:
And that's still a fairly locked down shot. Add any type of serious movement (handheld, jib, dolly, steadicam) and the flicker is going to get worse. Hopefully someday we'll be able to lock two DSLR's together so their shutters are in sync, but until then, this technique is going to be extremely hit or miss.
Click to expand...

I think the best way to demonstrate this will be to shoot footy, process it, and see how bad it gets. Certainly some situations will make it unusable. Why don't we do some test shoots and validate the phenomenon. It will be interesting to see where the "ceiling" is. Thanks for the input.
 
Rick Burnett said:
Very Cool! I enjoyed it. I've done some experiments with single sensor HDR techniques using consecutive frames, but motion was always an issue. 2 cameras seems like a good way to go, if only there were small enough cameras that you could sync the shutters.
Click to expand...

No question 2 will have less motion problems than 1 (with consecutive frames.) Truly as noted, shutter (without sync) will be a problem. How bad? What's the method for minimizing when you CAN'T sync the shutters. Good question. (As already noted.)
 
Andrew McCarrick said:
In the long, I really don't see a use for this as a regular tool. It's great if you're going for the hyper-real look for some shots. But HDR to retain highlights and grade it back to a normal look; why not just go for a better camera? A beam splitter plus 2 DSLRs will likely be 8 to 12 grand; that's borderline F3, C300, Scarlet territory.
Click to expand...

My first DIY beam-splitter cost less than 40usd. I'm going to upgrade the design and it will cost around 100usd. The first camera that I tested (because it shoots 24mm wide and resolves better than a DSLR on those shots) was the HX9.

HX9 costs less than 400usd. Grand total of less than $1K
T2i costs less than 500usd. Grand total less than $1.2K
NEX5N costs less than 600usd. Grand total of less than $1.5K
GH2 less than 900usd. Grand total less than $2K

The closest isn't within 11K.

That's why this might be fun to REFINE. It's certainly NOT there now, but, this was my first attempt using this workflow. What if it gets better? :) Wouldn't that be fun? Wouldn't that be interesting?

"Hyper-real"
Hmmmm....interesting. Does the final shot at 0:17 look "hyper-real" to you? I was actually quite surprised how natural it looked. In fact, the only part that I didn't like, (as part of the curtain), after further inspection was EXACTLY what was in the image....in other words, it wasn't the processing at all ....it was a weird aspect of the image its self.

For me, the idea of spending 100 bucks and taking some time to work with the NEX5Ns that I have to gain that much more range is CERTAINLY worth it. I understand that you'd rather spend the extra 12K and buy a cam that natively has 14 stops. (Not saying this method will produce 14 stops) Fair enough. That's a valid opinion. I just wanted to make sure to mention the fact that this can be done with ***MUCH LESS MONEY*** than any/either C300, F3, Scarlet. Cheers!
 
luke stewart said:
Looks awesome man, keep experimenting! What happens if someone walks through the shot though? Do you have to roto around them?
Click to expand...

Indeed, if you're going to use 1 camera (that doesn't have some sort of HDR built in, like the hacked T2i), then motion is a problem. Also, we'll see how obvious "non" synced shutter looks. (As I mentioned, the stutter in this image is the result of the shutter speed being auto set too high by the camera when setting the exposure compensation to -3.) So if folks are walking through your frame, you cannot use one camera.

The way you use 2 is with a beam-splitter. Then motion problems like folks moving around isn't a problem (not that stutter won't be...but we'll see how bad it gets.)

Cheers! Glad you like where it's heading. Who knows how good we can make it? We'll find out!
 
HHL said:
My first DIY beam-splitter cost less than 40usd. I'm going to upgrade the design and it will cost around 100usd. The first camera that I tested (because it shoots 24mm wide and resolves better than a DSLR on those shots) was the HX9.

HX9 costs less than 400usd. Grand total of less than $1K
T2i costs less than 500usd. Grand total less than $1.2K
NEX5N costs less than 600usd. Grand total of less than $1.5K
GH2 less than 900usd. Grand total less than $2K

The closest isn't within 11K.

That's why this might be fun to REFINE. It's certainly NOT there now, but, this was my first attempt using this workflow. What if it gets better? :) Wouldn't that be fun? Wouldn't that be interesting?

"Hyper-real"
Hmmmm....interesting. Does the final shot at 0:17 look "hyper-real" to you? I was actually quite surprised how natural it looked. In fact, the only part that I didn't like, (as part of the curtain), after further inspection was EXACTLY what was in the image....in other words, it wasn't the processing at all ....it was a weird aspect of the image its self.

For me, the idea of spending 100 bucks and taking some time to work with the NEX5Ns that I have to gain that much more range is CERTAINLY worth it. I understand that you'd rather spend the extra 12K and buy a cam that natively has 14 stops. (Not saying this method will produce 14 stops) Fair enough. That's a valid opinion. I just wanted to make sure to mention the fact that this can be done with ***MUCH LESS MONEY*** than any/either C300, F3, Scarlet. Cheers!
Click to expand...

I was referring to a full rig like the Bullet. Not just the splitter. But I guess with some cheap pieces from other rigs you could potentially rig something together for under $4k or so.

And as far as the hyper-real comment are you reffering to your video or the other one posted, because I was refferring to the other one. Yours does look fairly good. The other video posted looks almost like a video game.
 
Andrew McCarrick said:
I was referring to a full rig like the Bullet. Not just the splitter. But I guess with some cheap pieces from other rigs you could potentially rig something together for under $4k or so.

And as far as the hyper-real comment are you reffering to your video or the other one posted, because I was refferring to the other one. Yours does look fairly good. The other video posted looks almost like a video game.
Click to expand...

Indeed Andrew. A full rig like the Bullet are very costly. If that were the only option, I would have to agree that the C300/Scarlet/F3 are much better options. (If only for size constraints alone!)

Also true that the other video looks hyper-real. I'm not a fan of that look at all. I'm more of a "I didn't know that was HDR video" fan. In fact I don't even care for the HDR moniker on this type of video. I'd rather call it the "Hey how about an extra two stops of DR" video. (But that isn't nearly as catchy, LOL!) Cheers Andrew. Good points indeed.
 
awelgraven said:
I'm interested in what technique you used for combining your two shots
Click to expand...
I can't wait to share it! :) Its so simple. That said, I'd like to VALIDATE that it actually works. I need to shoot some better quality footy that is more dynamic. I also see to see how much the resolution is affected by it. I can't wait until the footy starts rolling in from around the community!!! (LOL) No one has sent me any (which is fine). Looks like it might end up being a lonely path for a while. (No problem....just a fact.) I'll test it, then I'll post it! Cheers!
 
These guys made a HDR video from a beam splitter 3d rig.

But I don't like that the shutter speeds are different. If I were doing this I would attach a mild screw-on ND filter to one lens, allowing everything else to be consistant.

 
Ryan-Guy said:
These guys made a HDR video from a beam splitter 3d rig.

But I don't like that the shutter speeds are different. If I were doing this I would attach a mild screw-on ND filter to one lens, allowing everything else to be consistant.

Click to expand...

Funny you should mention this, I was just thinking the same thing and wondering why no one has been using ND so they can keep the shutters the same speed.
 
Rick Burnett said:
Funny you should mention this, I was just thinking the same thing and wondering why no one has been using ND so they can keep the shutters the same speed.
Click to expand...

That's how I would do it, lol! The only reason that I didn't is because the ND that I have tends to have a color cast. I didn't want people to think that the image processing gave it that color cast. I totally think that is the way to do it.

To go a step further, you could actually use an ND as the "mirror" (4x4)...that way you get the ND effect for the camera that captures the highlights. You'd get the mirror effect for the camera that captures the low lights. Good point.
 
You must log in or register to reply here.
Back
Top