Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Dynamic Range comparison of RAW v/s video mode on an HDSLR  (Read 3301 times)
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« on: July 29, 2011, 11:24:00 AM »
ReplyReply

Hi
I didn't know this till today, but the dynamic range of an HDSLR (I tested my 550D) in RAW mode is possibly 3-4 stops greater than in video mode. I wonder why they don't mention that anywhere - didn't see it in the Zacuto tests either.

Is there anybody else with the same kind of experience? If yes, why do you think the dynamic range is lower?

In case you want to see the results of my crude test, here it is: http://sareesh.com/2011/07/dynamic-range-comparison-of-raw-vs-video-mode-on-the-canon-550d/
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
fredjeang
Guest
« Reply #1 on: July 29, 2011, 12:48:14 PM »
ReplyReply

That's why we, photographers, are so concern about raw video. I move the cursors with Red files and it doesn't fall appart. Maybe not as huge as what we are used to in photography because I think that Red Raw still uses a sort of compression, but it's raw and it changes all.

About DR, that is linked to isos, there is this interesting link in Red users forum: http://reduser.net/forum/showthread.php?12536-Understanding-ISO-with-the-RED-ONE.

Ah Sareesh, out of topic, I'm starting the Nuke learning and will take some training with pros in sept because there is too much to know. I'm just getting used of the interface. Very good! Do you know how is the DPX workflow with Nuke, can it be setted automatically in the project settings?

and my other question is: Would you recommend to convert the edited footage to images sequences as a rule for Nuke?

Thanks.
« Last Edit: July 29, 2011, 12:58:46 PM by fredjeang » Logged
UlfKrentz
Sr. Member
****
Offline Offline

Posts: 308


WWW
« Reply #2 on: July 29, 2011, 03:03:06 PM »
ReplyReply

Hi Sareesh,

did you try the technicolor cinestyle mode with the canon? I donīt know how this effects the 550D, but it should be applicable and it is pushing boundaries of dynamic range to a complete different level.

Cheer, Ulf
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #3 on: July 29, 2011, 11:46:10 PM »
ReplyReply

Nuke works very well with DPX (and even openEXR or any other image sequence). If you understand the way 'channels' work in Nuke, everything else is a breeze. You probably know this already, but before you train with the pros, may I suggest going through the user manual, then watching and practicing all the tutorials that can be downloaded from their website.

Regarding your second question: My highly opinionated answer is: Yes. I would convert the natively edited footage to image sequences prior to any image manipulation (both spacial and temporal). My favorite is working in 32-bit TIFF for super-heavy work, or 16-bit TIFF for the average job. All uncompressed, of course.

Images sequences are easier and faster for computers to understand and read - unlike chunks of data (like avi or mov wrappers around compressed data streams). The only downside is the amount of hard disk space and the size of RAM - both of which are cheaper than CPU and GPUs. Image sequences also render easily, and if the system crashes during the render, you only have to start from where it crashed - and not from the beginning.
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #4 on: July 29, 2011, 11:57:08 PM »
ReplyReply

I haven't used cinestyle. Maybe I should give it a try.
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
fredjeang
Guest
« Reply #5 on: July 30, 2011, 05:20:33 AM »
ReplyReply

Nuke works very well with DPX (and even openEXR or any other image sequence). If you understand the way 'channels' work in Nuke, everything else is a breeze. You probably know this already, but before you train with the pros, may I suggest going through the user manual, then watching and practicing all the tutorials that can be downloaded from their website.

Regarding your second question: My highly opinionated answer is: Yes. I would convert the natively edited footage to image sequences prior to any image manipulation (both spacial and temporal). My favorite is working in 32-bit TIFF for super-heavy work, or 16-bit TIFF for the average job. All uncompressed, of course.

Images sequences are easier and faster for computers to understand and read - unlike chunks of data (like avi or mov wrappers around compressed data streams). The only downside is the amount of hard disk space and the size of RAM - both of which are cheaper than CPU and GPUs. Image sequences also render easily, and if the system crashes during the render, you only have to start from where it crashed - and not from the beginning.

Thanks Sareesh,

Yes, I'm actually starting to watch the trainings provided by the Foundry.

The good thing is that on the few europeans official training centers recognised by the Foundry, 2 are actually in Madrid. Nuke has been popular in Spain since the begining so the access for official or unofficial training is easy in the spanish capital and that's important for me for time-efficiency reasons.

I like very much the feeling, the interface.

Cheers.
Logged
ChristopherBarrett
Guest
« Reply #6 on: August 09, 2011, 11:56:23 AM »
ReplyReply

Just to blow your minds a little more... stills out of the Red One versus the P65+.  Check the window detail in the Red file.  

<Right Click><Open Image in new window> for fullscreen-ish

Red One / Cook Panchro/i 18mm


Arca Rm3d / P65+ / Schneider 35mm XL


Very....... interesting.  You can't get nearly the detail out of the Red... but ohhhh the tonal range.
« Last Edit: August 09, 2011, 11:58:02 AM by CBarrett » Logged
fredjeang
Guest
« Reply #7 on: August 09, 2011, 02:56:50 PM »
ReplyReply

Interesting, today I just fall on a quick Nuke tutorial that precisely talks about that.

Let's assume that the footage is in linear colorspace,
When you pass the mouse over the white that appears really without info, you notice that the RGB values where in fact moving and telling that there were hiddeen informations on those blowned highlights,

Actually the guy in Nuke, just with the color correct tools, recuperated a full information, exactly as you would see in the stills version, and it wasn't Raw.

I'm sure that Resolve would do wonders because it's highly specialized.

I think that those high-end softwares are really key in the chain. (I saw a guy in Nuke actually erased a motorcycle and road sign on a moving footage in less than 1 minute with zero painfull procedure and you couldn't see anything at all).

I think that we should consider those post-prod softwares fully part of the set, in the sense that they are almost like part of the cameras performance. I mean the recuperated footage was white, completly blowned, but there where in fact infos in it that could be recuperate in a few seconds.

And also, I'm not far from thinking that the Cooter's PS use in motion is not a bad idea at all. Using scripts (I tried it today) we can automatize all the color corrections frame by frame. Honestly, when Coot came with this in the forum I was kind of sceptical but now I'm less and less and it's a worth path to explore (and explode the HDs). Just treat one image as a still in PS and apply with scripts all the circus in batch. As we are in low res for PS it's pretty fast.

Let's work, let's work. We'll get it !


ps for Chris: But just a question, this blown window could it be recuperated in Red CineX ?

ps2 for Chris erased ps 1 : Chris, I think I get it all wrong and things are reversed, it's late after a long day, could it be that the blowned windows was the P65 ? so the Red DR is impressive. (my mistake came that the Arca Rm3D is almost like R3D and I was sure that the bottom was the Red...mmm I need some rest)
« Last Edit: August 09, 2011, 03:32:33 PM by fredjeang » Logged
ChristopherBarrett
Guest
« Reply #8 on: August 09, 2011, 03:30:30 PM »
ReplyReply

Yes, the blown window is the P65+ and I've got hilight recovery set to 30. Interesting.
Logged
fredjeang
Guest
« Reply #9 on: August 09, 2011, 03:35:30 PM »
ReplyReply

and the Red was right-out the box or in fact the original was like the P65 and you recuperated in CineX ?

But even recuperated or not, this is, as you say, VERY interesting...

It means that if Red had high reso still capability, well, the choice falls by itself.
Logged
ChristopherBarrett
Guest
« Reply #10 on: August 09, 2011, 04:08:37 PM »
ReplyReply

You have several color sciences (or base profiles) available on the Red.  I used Red Gamma 2 as it is the preferred base point for many applications.  There is also RedLog Film which yields even more detail in hilights and shadows, but that profile takes a lot of grading to get a pleasing image because it starts so flat.  The hilight detail was there in the beginning.  Nothing required recovery.  The Red grade has had a lot of work done to match the contrast of the P65+ image.  I pulled down the shadows, pushed the midtones, brightened the hilights just a touch so they felt more natural and then dropped a curve on top of it to increase contrast in the midtones.  It wasn't as pleasing "out of the box" as the P65+ image but it did get there and retained the hilights in the process.

I only spent an hour or so on this test and perhaps I could be more clinical about it, but for me, this was a real world approach.  This all prompts me to begin more testing.  The next thing I really want to do is approach an architectural assignment with an Epic, Canon lens mount and my TS-E lenses.

CB
Logged
tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #11 on: August 09, 2011, 04:39:03 PM »
ReplyReply

Yes, the blown window is the P65+ and I've got hilight recovery set to 30. Interesting.
how does it look like when you apply the "linear film curve" to the p65 file?
Logged
ChristopherBarrett
Guest
« Reply #12 on: August 09, 2011, 06:33:23 PM »
ReplyReply

Here is a P65+ shot that's what I'd call a stop underexposed, set to Linear and pulled in various directions to try and get there.  Much more tone in the window (I really like the window here), but not real pretty overall.





« Last Edit: August 09, 2011, 06:35:18 PM by CBarrett » Logged
tho_mas
Sr. Member
****
Offline Offline

Posts: 1696


« Reply #13 on: August 10, 2011, 02:32:28 AM »
ReplyReply

Here is a P65+ shot that's what I'd call a stop underexposed, set to Linear and pulled in various directions to try and get there.  Much more tone in the window (I really like the window here), but not real pretty overall.
yes, due to the linear film curve it looks somewhat flat. But the linear film curve reveals highlight differentiation that was actually captured whereas the standard film curve is quite steep in the highlights. However the standard film curve looks much better overall.
I do use the linear curve from time to time to "recover" highlights... but I mostly blend in the highlights from that "linear" TIF in Photoshop on layers (so actually a kind of HDR workflow). Likewise I prefer the "extra shadow" film curve over C1's HDR tool to open up the shadows.
Logged
Sareesh Sudhakaran
Sr. Member
****
Offline Offline

Posts: 547


WWW
« Reply #14 on: August 10, 2011, 05:56:53 AM »
ReplyReply

Very soon we'll have a medium format video camera that can do everything!
Logged

Get the Free Comprehensive Guide to Rigging ANY Camera - one guide to rig them all - DSLRs to the Arri Alexa.
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad