-
HDR Processing with Only 1 Take?
Hi CC,
I’ve been learning about HDR 32-bit processing and find it amazing. I thought color matching and luma matting was cool, this looks like the cat’s pajamas.
Is it possible to combine different exposures of 1 take of raw video and get the same results as several takes at different exposures? The reason why I ask is that beyond BG or landscape photography, this would be pretty much impossible trying to say video your dog as it wouldn’t match take-to-take.
With CS5 would one be able to get the same combined results using an in-software method?
Thanks for any feedback.
-
11 Replies
-
It wouldnt be the same result, but then again theres no way to do HDR videography unless you have a RED or something that can take simultaneous dual exposures. To do what your talking about would still require a RED or similar to get RAW video. You’d be better off shooting film for the latitude it would offer to attempt such a thing, but you can already achieve a similar effect using curves.
-
Yeah, I can’t afford film for this so I guess I should do more investigation in RED and other simultaneous RAW video formats.
I learned the SLDR cameras are supposed to provide the same resolution, but I haven’t heard of any that take simultaneous exposures.
Like you said, worst case scenario I can use levels/curves as long as I have 32-bit resolution.
Thanks David!
-
Um, that should be “i cant afford red so i’ll look into film”. Editing at 32bit wont make any difference if you didnt capture at 12bit or above (all consumer/prosumer cameras capture at 8bit, which is compressed playback depth). I think you mean video DSLR’s, which might sample the sensor at a higher resolution, but still only offer 8bit. Your going to need at least a six figure budget to get what your looking for. That or just shoot all timelapse.
-
Yeah, my bad, DSLR.
I just love the HDR look this person put together here with two Canon 5Ds and a beam splitter:
I hear their sound is crap, but this other guy shot an award-winning documentary with the same camera thanks to a custom stereo out mod and a Sennheiser boom mic:
https://www.hellandbackagain.com/index.htm
Not quite sure what you mean by $100,000’s. ? I know REDs are not cheap. But that doesn’t have to be reality. I’m really just investigating HDR process, that’s all.
Some contents or functionalities here are not available due to your cookie preferences!This happens because the functionality/content marked as “Vimeo framework” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.
-
[David Ghast] “Um, that should be “i cant afford red so i’ll look into film””
No, it correctly should be “I can’t afford film so I’ll look into Red.” Otherwise there would have been no market for Red at all. Instead they are transforming the industry.
HDR tonemaps more than 8 stops of latitude down to approximately 8 stops for the delivery codec. As David said the only way to get HDR in one take is to use a camera that records greater than 8 stops of latitude so you have something to tonemap. Those options are essentially film, red, or regular 5-8 stop video cameras with a beam splitter as in your example.
Red has just released the Epic which can do HDR in one take natively. Red cameras are also the only digital camera that shoots in raw format.
DSLRs do not compare to red or film cameras. If you like the look of the video you posted, and your delivery is web video, then get two dslrs and a beam splitter and forget red and film. If you are looking for more than a web video, high quality for mass distribution in theaters or bluray, then consider film or red. You can rent a red camera for a few hundred a day. But it’s not a point and shoot camera, you need a crew to run it, or at least an experienced DP, as with film.
Cory
-
What I find interesting is that in that video posted the guy only had a total latitude of 4 stops.
What I learned from a separate tutorial is that it’s important to have 32-bit float to be able to have deep luminance information beyond a simple 8-bit rez.
I am working with a professional DP and practically genius 1st camera. However, still…Those red cameras are…Wow. To be taken seriously for sure. So I’m going to investigate not only f-stop latitudes but the latitudes of my entire operation before making that decision.
The thing is I really want to go the mile with this one if I’m talking about “beam splitters”. It’s a bummer that a high-end DSLR can’t provide that high of a rez for broadcast level output. I would think HD is HD. ?
I’m not totally scared off from renting a Red, but like I said, I will have to figure out what I’ve got to work with.
Thanks Cory!
-
[Max Jackson] “What I find interesting is that in that video posted the guy only had a total latitude of 4 stops.
Not quite. The canons have 6-9 stops of latitude, depending on who measures them (naturally questionable), lets say 8 stops. They used two cameras in order to stagger those 8 stops; two above normal, two below. Thus they came out with 12 stops of dynamic range, then tone mapped it down to ~8 stops for delivery.
The Red One will give you 11 stops of dynamic range natively. Film can give you 11-15+ typically. Red Epic in HDRx mode can give you 18.
What I learned from a separate tutorial is that it’s important to have 32-bit float to be able to have deep luminance information beyond a simple 8-bit rez.
Your delivery codec is going to be 8-bit, and thus ~8 stops. During render, the DR will be either compressed or clipped into this window.
If you work in an 8-bit color space in AE, then AE will automatically throw away color information before you can even work on it. By working in a 32-bit space, you have all the color information to work with, and can control what is thrown away and how (by using the color grading tools) when exporting to an 8 bit delivery codec.
As David said, if your source footage is 8 bit, as from a DSLR, working in 32-bit usually doesn’t do much for you. However if you are working with a 12 or 14-bit red raw codec in an 8-bit space, you’re wasting the time and money you spent on Red.
It’s a bummer that a high-end DSLR can’t provide that high of a rez for broadcast level output. I would think HD is HD. ?
“High end DSLRs” are high end still cameras that happen to take video. The professional 1dm4 is a great, professional still camera, but it’s a consumer class video camera. Their effective resolution is about 500×500 pixels delivered with hardware aliasing into a video codec set at 1920x1080p. HD is more of a marketing term for them.
Sony and Panasonic make real HD cameras that can deliver truer 1080p (measured at >800 lines of effective resolution) at ~8-10 stops of DR.
Red One blows away HD, delivering 4k (4,096×2,304). Red Epic delivers 5k (5,120×2,700 – 32% more).
You would pick red for either high dynamic range, or resolution, or those plus shallow depth of field, or the raw codec.
Cory
-
Ohhhkay, it’s all starting to become more and more clear. It’s just that these “HD” definitions are rather blurry considering one source is different than another.
That’s a lot of stops considering there’s only 15 in the Ansel Adams Zone System. It’s really amazing how adjustments to 32-bit content can pull values out of pure white.
So DSLR cameras can provide sufficient HDR processing, the dynamic portion being their extreme latitude for exposure. However, their camera resolution is much broader than their video. So like you say, if you want to go larger than 1080p, go RED cuz there’s not any other options really. Right? Am I getting warm?
The more I read about RED cameras, the more amazing it sounds. I’m thinking I should take a class and then maybe rent one and get use to it’s settings and response before hand. It’s about $300-400 a weekend rental which is about as much as renting a bowling alley for a birthday party, but they’re really cool. I’ll have to look into RAW format more in terms of codec specs for transfer.
Crap, that and think about my monitor set-up. Sheesh.
Thanks Cody!
-
Beamsplitters cause discoloration in the splitted image, how is that going to help with HDR?
-
“Is it possible to combine different exposures of 1 take of raw video and get the same results as several takes at different exposures?”
NOT really HDR, but we’ve occasionally converted video to an image sequence in Photoshop, then opened two different RAW versions (overexposed and underexposed) in After Effects. Then we put the two versions into a comp with some motion masking or blend modes. Our goal is usually to rescue footage with too much backlighting (think of someone standing in a doorway with too much sunlight behind her.)
Seems like the same concept would work if you are starting with RAW image sequence. (There’s a tutorial here on the Cow about opening an image sequence in AE as RAW.) It’s not a workflow you’d want to use routinely, but accomplishes some of what you’re trying to do. It can be used with existing footage and a LOT cheaper than two RED cameras and beamsplitters.
I haven’t tried it, but there might be a way to get Photoshop to take two image-sequence files and Merge to HDR.
Jim
ColorBurst Video
Reply to this Discussion! Login or Sign Up