PDA

View Full Version : Dynamic range stress test



Frank Glencairn
10-08-2012, 05:46 PM
Shot some DR tests today. Nothing fancy or artsy, emphasis was on contrast handling
(not on smooth camera moves - had that lightweight travel tripod with the crappy head again - but walking several miles with a 40lbs Vinten?)

So I went out to the German version of Central Park at mean noon sun.
That, combined with a lot of trees is usual a receipt for DR disaster.
Deep shadows with puddles of super bright sunlight.

Point is, the material It looks not really spectacular, but just natural - more what your eyes see (not exactly but you get the idea).
So yeah, mission accomplished BMC.

Still not exactly happy, with my post workflow though.

Tech specs:
24p @ raw
Vintage Vivitar Series 1 Zoom
Crappy sticks and head
Converted and graded in Cineform.

For the record: Color spaces start driving me mad.
The only time it looks good is when I grade on my Dreamcolor monitor set to 709 and burn it on Bluray.
But with H264, it looks different all the time, from each compression program, on each monitor, with each player on each computer and on Youtube and Vimeo too.
I hope we can get this crap unified some day.


http://youtu.be/_DKQQCSVx0w

Frank

Matthew Bennett
10-08-2012, 06:07 PM
Any footage from Germany is beautiful to me.

nickjbedford
10-08-2012, 06:08 PM
I don't know if you've posted this information anywhere else, but what are your returns from using the CineForm RAW conversion?

It creates 12-bit wavelet compressed raw files, right?

Frank Glencairn
10-08-2012, 06:11 PM
It's a nice balance between a quick workflow without roundtrips, speed, quality and file size.
Also I really like those many options you can select for demosaicing and matrix.

Asladek
10-08-2012, 06:17 PM
Hey man,

Great test, it looks like it handles the range pretty well. Any chance
of you posting 1080 version in the future?

Frank Glencairn
10-08-2012, 06:20 PM
The clip is still processed by youtube. Full HD should be available any minute.

nickjbedford
10-08-2012, 06:22 PM
It's a nice balance between a quick workflow without roundtrips, speed, quality and file size.
Also I really like those many options you can select for demosaicing and matrix.

Sounds good. What size ratio are we talking? 2:1? 3:1?

Frank Glencairn
10-08-2012, 06:25 PM
3.5:1 ish.

nickjbedford
10-08-2012, 06:29 PM
3.5:1 ish.

Cool, thanks. Like I tweeted, the range is fantastic. Usually what you expect out of 12+ stop cameras shot with all that information stored in a raw file and all that information pulled back out of the default tone curves.

More often than not, simply shooting raw makes camera's range far more impressive regardless of the rated dynamic range because the tone curve isn't applied destructively.

Frank Glencairn
10-08-2012, 06:32 PM
Yes, on all other cameras under 18000 bucks, most of the lights would be completely blown out and the shadows just dark blotches. This is RED HDR and Alexa territory IMHO.

nickjbedford
10-08-2012, 06:34 PM
Yes, on all other cameras under 18000 bucks, most of the lights would be completely blown out and the shadows just dark blotches. This is RED HDR and Alexa territory IMHO.

It's interesting isn't it. More often than not, I can pull HDR level dynamic range out of a single Canon raw file simply because the "norm" is apparently about 8 stops, not the full range in the raw file. DSLRs record far more range than a simple JPEG straight out of camera portrays. You just have to dig in the raw file for it!

funwithstuff
10-08-2012, 06:41 PM
Very nice indeed. Beautiful images, and thanks for sharing.

Cedric Akins
10-08-2012, 06:44 PM
Thanks Frank.

CineMac
10-08-2012, 07:25 PM
Ah nice, beautiful... Gah! The wait.... is... excruciating... :-)

Thanks again Frank! You break out those Zebra lenses yet?

Ha ha, gotta keep asking, those things just look so cool.

PaPa
10-08-2012, 07:32 PM
man, some of those shots simply blow me away with the level of detail in the highs and lows. Absolutely amazing. Very exciting indeed. Thank you for your test!

taubkin
10-08-2012, 08:13 PM
Very interesting as usual Frank. Looks a lot like a 16mm shot, but with better sharpness. There is some magenta in the highlights, but since that is a solved issue I take it was an aesthetical choice, or just my monitor being off.

Now that's noon in Germany? Looked like beautiful winter afternoon to my tropical standards... :)

Andrew
10-08-2012, 10:00 PM
Wow Frank! I am blown away by this footage and how great it looks even in such high contrast situations! Great color! Thanks for sharing.

Hybaj
10-09-2012, 12:50 AM
The most filmic grade on BMC footage yet. While it's not a block-buster type look it's absolutely perfect for a "sunday" after lunch family movie or documentary work. Good job Frank and kudos to BMC for allowing this to happen.

Gwangjuboy
10-09-2012, 03:23 AM
With respect and not to be rude but probably the last thing the BMC needs testing on is DR. But trying to get someone to test IS on the kind of standard EF lenses including zooms most people will use is like getting blood out of a stone.

Fabián Matas
10-09-2012, 03:31 AM
Ei Frank, nice work How did you frame "in camera" for the 2'35:1? About HDR, I'm not sure but...maybe is possible doing It via photoshop, slow but possible at the moment.

Andi1
10-09-2012, 03:32 AM
well i liked the test a lot.. but an IS test would be great!!

Frank Glencairn
10-09-2012, 03:38 AM
Sorry, I don't have IS glass, since I don't own Cannon lenses, but I think Mark already have done IS tests.
There is not much to test with IS anyway.
They do what they do on every camera, it's not that they behave different on a BMC - it just provides the juice. Stabilization is don by the lens.

Framing was done with an external monitor with guides.

Just In case someone wants to download the clip: https://vimeo.com/51044821

Gwangjuboy
10-09-2012, 03:39 AM
But surely all the footage so far is ultimately a test for DR. I have no problem with the footage, it looks good. I just don't know why the trouble is being made to test something I don't think anyone doubts on the camera.

I would say the biggest thing is movement, IS, depth of field, distortion of the Wide EF lenses are obvious tests but I confident we won't see anyone do them until the masses get the cam. why? I've no idea.

Frank Glencairn
10-09-2012, 03:47 AM
I'm doing a docu next week, so I wanted to make sure to know how much I can push the BMC - hence the test.
I take my time to get a feeling for each camera I work with, so I have confidence, that I expose exactly right every time - less stress on set and in post :)

Gwangjuboy
10-09-2012, 03:58 AM
I'm doing a docu next week, so I wanted to make sure to know how much I can push the BMC - hence the test.
I take my time to get a feeling for each camera I work with, so I have confidence, that I expose exactly right every time - less stress on set and in post :)


Hey Frank I just wanted to make clear I appreciate you getting some footage out there. It's just a frustration I have to see the kind of lenses I would use and the IS working with them. Your footage is actually very nice, makes me want to visit a friend I've got in Germany.

BM4EVER
10-09-2012, 04:19 AM
Yeah he is. This is a great test; noon is a nightmare for most cameras and the BMCC handled it with aplomb. Thanks Frank, very valuable to know that we're less limited in the times we can shoot outdoors now to get lovely shots :)

RyGuy
10-09-2012, 04:46 AM
Great test Frank. Thank you. Can you tell us what you did in the grade? It'd be nice to know how much massaging the images need to make it look that good.

MiguelFranco
10-09-2012, 05:25 AM
Thanks for the nice footage Frank. Would it be too much to ask for a simillar test using Prores? That lens seems sweet, did you use an ND in any situation?

Jim-Box
10-09-2012, 05:36 AM
I like a lot Color and DR. But I see too much ... Flickering? or choppy play. I dont like the motion. Panning shots and People walking are not smooth. I see a problem there.

nickjbedford
10-09-2012, 06:44 AM
Huh? The motion is typical of 24fps at 180 degree shutter.

t.p.
10-09-2012, 07:07 AM
great footage , thanks for the test frank

Brian@202020
10-09-2012, 07:10 AM
Huh? The motion is typical of 24fps at 180 degree shutter.

This is true, but there are some cases where it makes sense to venture outside that box. I've been known to increase the shutter angle a little if a faster pan is required and then go back to 180 degree after. The trick is to find the acceptable middle ground in such a scenario. You don't want to increase the angle too much otherwise it will stand out in the edit next to the 180 degree shots. At the same time staying stuck in the 180 degree box with every shot limits your possibilities. 180 degrees is more of a guideline than a rule, and I hope eventually the BMCC has every degree shutter angle from 1 to 360.

Frank Glencairn
10-09-2012, 08:09 AM
It's pretty smooth with the original files. Actually it judders less than a FS100 at 25p and the motion rendering is very nice.
Most of the problem you see where introduced with the h264 encoding, maybe I need a higher bitrate or the codec has a hard time to handle the level of detail.

Frank

AndrewDeme
10-09-2012, 08:32 AM
It's pretty smooth with the original files. Actually it judders less than a FS100 at 25p and the motion rendering is very nice.
Most of the problem you see where introduced with the h264 encoding, maybe I need a higher bitrate or the codec has a hard time to handle the level of detail.

Frank

Yep, agree....higher dynamic range = more information and low data rate codecs struggle to keep up.

Time will tell but we now need fibre to the home and larger hard drives and the cycle continues.......

Brad Ferrell
10-09-2012, 10:06 AM
"had that lightweight travel tripod with the crappy head again " - this explains the judders in the pans. :)

I do prefer the 25P in a wierd way, at least for an ultra low-budget ($25,000).

Kholi
10-09-2012, 11:07 AM
Dunno what's wrong with me, but when I saw the thread title I was like "Finally, someone shot a DR chart"

>.<

Was actually excited to see what the camera was actually measuring for DR so we could use some real numbers. Weird.

Frank Glencairn
10-09-2012, 11:22 AM
I'm not a huge fan of charts - I use them to some degree, but real world footy is what counts at the end.
A chart is always finite and defined (of course) but the real world has infinite graduations, colors, textures, detail.
So there is always a huge difference.
Have you seen the charts John posted? Optimo and Hawk anamorphics.
If you look at the chart you would have to say "throw that crappy glass away". But what those charts show, doesn't mater if you go and shoot something.

Regarding the judder:

There are two sorts. One is crappy tripod motion (mostly panning) and the second one was introduced by encoding for web.
Like the tilt in the first pan is pretty smooth on the original material, but looks horrible after encoding.

I promise, next time I break out the heavy duty Vinten.

Frank

Brian@202020
10-09-2012, 11:32 AM
I think both real world work AND charts are good myself. I, like Kholi, wish we could see charts from someone.

Dafilman21
10-09-2012, 12:08 PM
Thanks Frank, I am just happy to see new material being shot to help pass the time.

Matthew Bennett
10-09-2012, 02:45 PM
I think both real world work AND charts are good myself. I, like Kholi, wish we could see charts from someone.

I'm sure Abelcine will be on this as soon as they lay hands on the camera...

Kholi
10-09-2012, 04:06 PM
If you look at the chart you would have to say "throw that crappy glass away". But what those charts show, doesn't mater if you go and shoot something.

Frank

A year ago I would've agreed with you. As of now, there's nothing anyone's shooting that tells me anything that I don't already know. A chart or a wedge would tell me something new. While I don't think that Blackmagic's advertising 13 stops and not getting a usable 13, I would like to know what the number is exactly.

Personally, I'm well aware of the differences between real world and charts, doesn't really matter because the charts still influence what you shoot.

Never was a chart guy... until this camera came 'round.

CoolColJ
10-10-2012, 03:26 AM
Really nice colours on this camera

nickjbedford
10-10-2012, 07:32 AM
Why do I stress the need for a raw camera?

Top: Default settings in Lightroom. Not unlike the results you get when shooting a DSLR in video mode (though way sharper here).
Bottom: 5D Mark III raw file used to its full potential.

http://www.nickbedford.com/uploads/asaasdads.jpg

hmcindie
10-10-2012, 08:51 AM
That jpeg also has stuff hidden. Here's my version from just the upper jpeg. As you see, there was a lot of shadow detail that I could still extract from the 8-bit jpeg.

1292

razz16mm
10-10-2012, 09:00 AM
I spent some years producing video programming and still photography for an art museum. Charts were a critically important part of the process for controlling accurate color and tonal reproduction of the art works.
I use charts as a neutral reference point for understanding how a camera or film stock performs so I can get the most from it. I calibrate my cameras to my light meter with charts , light and expose to the meter, even with video cameras. I calibrate monitors and printers for accurate reference and results using charts.
If one knows how to use them, charts are an invaluable tool for improving your photography.

CaptainHook
10-10-2012, 04:06 PM
Why do I stress the need for a raw camera?
You could get very similar with LOG and a good codec, and still not too bad even with a more 'video' profile. I love the latitude in pushing raws around too, but i would much rather get it in camera and have to do very little post. If you can do that, raw isn't 'needed'. But yes it would be nice.

CaptainHook
10-10-2012, 04:25 PM
hmcindie's post made me curious so i had a '1 minute play' with the top jpg too. Sure there's banding in the sky, but it's not bad and neat video would prob sort it if i felt the need to go this far (i dont think i would go as far as the raw example unless the scene really called for a low-con/hdr-ish kind of look). Would prob use a pola and/or grad, and maybe some diffusion if i wanted this low contrast look 'in real life' so wouldn't have to push so far in post too.

http://f.cl.ly/items/1U1E2a2t3H3Q3R3D2v3T/hook_quicktest.jpg

nickjbedford
10-10-2012, 04:33 PM
Yeah I'm not saying there's no room in JPEGs, but you can see that detail and range is lost when using a "baked" source. Maybe I'm just too used to the freedom that shooting raw stills has given me.

CaptainHook
10-10-2012, 08:32 PM
I guess i'm trying to say in my experience it's generally a 'better' result to "get it" at source and to focus there if possible. There's a few industries (music production a prime example) where because post options have drastically reduced in price there's a tendency to record 'flat/neutral' and create the sound/look in the mix/post since it's so accessible to everyone now. Even worse it's expected that performances will be fixed/enhanced in post. Things tend to sound/look better to me when people make decisions and commit upfront on input. BUT, i don't always work that way myself either since it really depends on the job. Flexibility after-the-fact is great until it starts to become a crutch or the famous "we'll fix it in the mix/edit/grade/whatever". :P I'm not gonna go start shooting JPG's on my mk3 either and for some of our work, we DO rely on what we can do with RAW. But i don't like the idea i NEED raw to create an image the way i want as i'd rather be able to get it on capture. Shooting on my rz67 pro II is a blast, but with the cost of each frame the way it is now pressing the shutter is a lot more 'considered' - which i enjoy. I'm being overly philosophical about it tho. :P

Would a good LOG file not satisfy you?

nickjbedford
10-10-2012, 09:49 PM
I guess i'm trying to say in my experience it's generally a 'better' result to "get it" at source and to focus there if possible. There's a few industries (music production a prime example) where because post options have drastically reduced in price there's a tendency to record 'flat/neutral' and create the sound/look in the mix/post since it's so accessible to everyone now. Even worse it's expected that performances will be fixed/enhanced in post. Things tend to sound/look better to me when people make decisions and commit upfront on input. BUT, i don't always work that way myself either since it really depends on the job. Flexibility after-the-fact is great until it starts to become a crutch or the famous "we'll fix it in the mix/edit/grade/whatever". :P I'm not gonna go start shooting JPG's on my mk3 either and for some of our work, we DO rely on what we can do with RAW. But i don't like the idea i NEED raw to create an image the way i want as i'd rather be able to get it on capture. Shooting on my rz67 pro II is a blast, but with the cost of each frame the way it is now pressing the shutter is a lot more 'considered' - which i enjoy. I'm being overly philosophical about it tho. :P

Would a good LOG file not satisfy you?

A good LOG file will very much satisfy me. What I'm saying is that it's nice to say, "Go shoot on your DSLR", but then I say, "My DSLR bakes a look in that I otherwise want control over". I'm totally cool with shooting either ProRes film log or raw on the BMDCC.

razz16mm
10-11-2012, 12:48 AM
Capturing an image with the DR compressed into 7 stops of video range via a log curve is not the same thing as full linear capture of the DR in a raw file. You still lose control of fine adjustment range in deep shadows and in highlights.

nickjbedford
10-11-2012, 12:53 AM
Capturing an image with the DR compressed into 7 stops of video range via a log curve is not the same thing as full linear capture of the DR in a raw file. You still lose control of fine adjustment range in deep shadows and in highlights.

Well to be fair, I'd take a log image in 10-bit minimum (like the BMCC).

CaptainHook
10-11-2012, 05:19 PM
What I'm saying is that it's nice to say, "Go shoot on your DSLR", but then I say, "My DSLR bakes a look in that I otherwise want control over". And i say "control the look in front of the camera, not in post". ;)
I've read other posts of yours saying you want to learn everything you can about cinematography, but these posts sound like you're more interested in being a colourist and working in post? Which is fine, and i understand wanting to do both, but IMHO if you shoot now with your mk3 and learn how to make it look great (since you can't rely on post you'll be forced to do it in capture), those lessons and experience will only help your BMCC work.

Anyway, i think the horse is hacked to pieces now. Hope the camera ships soon. Too many frustrated customers. :(

nickjbedford
10-11-2012, 06:41 PM
And i say "control the look in front of the camera, not in post". ;)
I've read other posts of yours saying you want to learn everything you can about cinematography, but these posts sound like you're more interested in being a colourist and working in post? Which is fine, and i understand wanting to do both, but IMHO if you shoot now with your mk3 and learn how to make it look great (since you can't rely on post you'll be forced to do it in capture), those lessons and experience will only help your BMCC work.

Anyway, i think the horse is hacked to pieces now. Hope the camera ships soon. Too many frustrated customers. :(

Don't get me wrong. Give me some lights, a set and a camera and I'll make it work to the best of my ability and the resources I have. I've been improvising for the most part in my photography since day one since I haven't had the big resources to really do photo shoots with strobes, set design and all that. I took this (http://www.flickr.com/photos/nickbedford/8008504475/in/photostream)with nothing but an iPhone 4S and some processing in the Snapseed app.

I do want to be the colourist for my own work, though. Not Canon (or some other colourist :p). It would be like me taking photos and getting someone else to develop them.

But it's all conjecture until we get the camera and produce work. I can say this and that, but it won't make any difference until I prove to myself that the camera was worth it. I think it absolutely will make a difference and we've seen just how much better the same material looks.

ari
10-12-2012, 12:32 PM
On a interesting aside, shooting on location during the day is all about DR. I think thats the really interesting thing to take away form this test, You spend such huge amount of time on location trying to even out the tremendous range of sunlight with HMI and diffusion, that it can actually become more difficult than shooting in a studio and creating controlled sunlight.
What I'm getting at is that for every extra stop of DR in camera you can rent a less powerful (and cheaper) HMI for outdoor work. Once you have the range in the DNG it's fairly easy to flag off parts of the image and create some depth in the DI.