how to gain lights to this video in post prod

Siby wrote on 1/21/2012, 9:29 PM
I recorded a video using Sony CX700V last week in 1080P, 24P

I had enough light in the stage. But I couldn't do much camera settings such as White balanceing, iris backlighht etc..

As you can see the dancers faces are dark in this video. Any suggestions how can I make this video better. Its looks dark now.



Comments

larry-peter wrote on 1/21/2012, 9:56 PM
Sony Fill Light is something you could look at, although you've got some nice rich colors in the clothing that will get washed out a bit, but it's an easy plugin to use.

Personally I would go with Color Curves. You have plenty of headroom in the white wall so you can take the handle on the top right node and move it upward to stretch your top levels upward a bit. Varying the length of the "handle" will adjust the range of levels affected. To get only the faces up a bit more, click on the curve around 1/3 of the way up from the bottom left to create a new node and move it upward slightly. Adjust the length of those handles to make the affected range wider or narrower.

If you still feel you need get the overall levels higher , add a Levels plug in and bring the "input end" slider to the left slightly. I suggest avoiding the gamma adjustment in either Levels or the Color Corrector. They're both heinous things. Get used to Color Curves.

Color Curves takes some experimentation to get comfortable with but it's well worth it. It can work near-miracles in tough lighting.
musicvid10 wrote on 1/21/2012, 10:04 PM
"Any suggestions how can I make this video better. Its looks dark now."

Sure.
In Sony Levels fx, set the following:
-- Input End: .900
-- Output Start: .040
-- Gamma: 1.25
This sets REC709 levels and emphasizes skin tones in underexposed video. It is underexposed because you used autoexposure against a white wall.

Set your render frame rate the same as your source, 24 fps
Reduce the right channel audio by -1.7dB.
Upload to Youtube again.
http://www.jazzythedog.com/testing/DNxHD/HD-Guide.aspx

redpaw wrote on 1/21/2012, 10:04 PM
You can try Frederic's exposure plugin - actually, it's on sale till Monday! only €29.90
check it out: http://www.fbmn-software.com/en/exposure.html
amendegw wrote on 1/22/2012, 4:29 AM
I can't add anything further to the excellent advice you are getting re: fixing backlighting, however, I'm seeing something in you footage that looks somewhat problematic.



If you shoot in 24p and render in 24p you shouldn't have this problem (remember there's both 24.000 fps & 23.976 fps render settings - don't mix the two). And make sure you use the "Match Media Settings" Wizard to set your Project Properties.

All that said, I've become a great proponent of using "Disable Resample" (the Vegas default is "Smart Resample") to prevent issues like the above. See: http://dl.dropbox.com/u/20447760/DisableResample-SonyCX700V.pngWhither Smart Resample[/link]

I've written a script that helps in this workflow: SetResample.zip

Hope this helps,
...Jerry

Edit: I just downloaded the YouTube Video. It's 29.97 fps. If you're shooting in 24p (assume 23.976??), and using "Smart Resample", I'm pretty sure that's the problem.

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

Frederic Baumann wrote on 1/22/2012, 6:27 AM
Hi,

Many thanks 'redpaw' for the recommendation!

Acutally the sale was only sent to selected contacts, and is not available on the web site.

However it is available from the FBmn Software facebook page. By the way feel free to 'like' the page to be notified of further special discounts.

Frederic - FBmn Software
farss wrote on 1/22/2012, 6:55 AM
I think your exposure problem has been as well dealt with as it can be by others.

One tip I got from someone whose directed more TV shows than I've had hot dinners:
"When shooting dancers try to keep their feet in shot."

From myself who shoots a few Indian dances, if you can wrangle a second camera use it to get some shots of just the feet.
Oh and also try to record a feed from the desk or better yet, get a copy of the original CD.

Bob.
Laurence wrote on 1/22/2012, 8:52 PM
Fredric, I wanted to look at your software and maybe take advantage of the half off offer but your web site is down.
Siby wrote on 1/23/2012, 3:42 AM
Hi Jerry and musicvid,
I made the changes you suggested and posted the videos. Not sure how to adjust the audio right channel as you said. so i didn't toched the audio.

The original video from the camera =>

Video with the changes =>

rendered with disabled resampling: =>

rendered as wmv w/o changing resampling =>

This looks much better. But I think there is something wrong with the render.
I posted the render details in the video description

Thanks for your valueable input. you guys are amazing how quickly you gave these tips. Very professionals.
Duncan H wrote on 1/23/2012, 4:30 AM
Hi Siby.

I've had a look at your various videos.

Clearly dramatic improvement from the first one you posted, with much more detail in teh faces.

I viewed the "Video with change" and I note that this is a 25 p mp4 (whereas your original was shot in 24p). There is still an image echo (frame issue) visible due to the disparity between 24 fps of the original and 25 fps of this version. It is particularly noticeable in the second half of dance. Again, suggest using "match project settings" in the render as options, as previously advised.

Your other version ("Render with disabled resampling" ) is a different format, being a WMV, although I note that this one is 24fps and perhaps unsurprisingly, you have dramatically improved the output as by matching framerates with the source footage, you've eliminated the image echo seen previously. I don't render to WMV , so you'll need other's advice, however, face with this same situation, I'd fire up the video plugin for colour curves, do some tweaking and re-render using this same 24 p setting and I think you might surprise yourself by the improvement in the finished video.

Alternatively, if I had your situation, I'd choose to render as an mp4, selecting say 720p and then use the "customise template" button immediately below the list and change the frame rate to 24 fps (film). I think you should then get very acceptable footage. Trust this helps.


Regards,

Duncan
amendegw wrote on 1/23/2012, 5:00 AM
Okay, Let's take this in order.

"Not sure how to adjust the audio right channel..."

I strongly suspect your problem is a symptom of the http://www.sonycreativesoftware.com/forums/ShowMessage.asp?ForumID=4&MessageID=777854Audio: 5.1 vs Stereo[/link] issue. Read that thread & it should answer all your questions.

Next, a MediaInfo report of a download of your original footage shows framerate of 23.976. The render footage is 25.000 fps. Here's how to fix that problem. First what the following YT tut to set your project properties:



Next you will need to modify your render template to 23.976 fps.



Now, you should be a master of the "good" method in musicvid's excellent YouTube procedure.

...Jerry

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

amendegw wrote on 1/23/2012, 6:23 AM
btw, back to your original question, "Any suggestions how can I make this video better. Its looks dark now." I spent 5 minutes chaining the FBmn Exposure Plugin & Color Curves. Here's what I came up with. I'm sure if I spent more time, we could make it much better.



Also, doesn't your CX700V shoot in 60p (50p?). There's so much motion in this video, the increased framerate would go a long way to smooth the motion.

...Jerry

PS: Those kids have a lot of talent!

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

craftech wrote on 1/23/2012, 7:05 AM
From the Camcorderinfo review of the HDR-CX700V:

Page 5

John
Siby wrote on 1/23/2012, 2:31 PM
I am very impressed with the results that shown with additional tools and color corrections by the professionals. First of I wanted to thank you to Jerry, John, Duncan, musicvid and others. I purchased FBmn Software Exposure GPU plug-in. Waiting for some feedbacks from Jerry. will post the new results shortly.

This was my first 24p footage. I guess no more 24p for these type of motion videos.
farss wrote on 1/23/2012, 3:32 PM
"This was my first 24p footage. I guess no more 24p for these type of motion videos."

If you're shooting under low light and that was low light shooting, 24p compared to 60p will gain you over 1 stop of light. I'd kind of favour the extra stop of light over anything to do with motion rendition myself, after all pretty well everything Hollywood does is at 24fps.

A couple of thoughts about this.
Firstly I'm not certain that this camera is really shooting 24p. Generally the only Sony cameras that shoot 24p are their expensive "CineAlta" cameras, the rest shoot 23.976. The difference is minute but when you're working in post getting it wrong can cause grief. Vegas supports both 24.000 (film) and 23.976 (IVTC Film).

Secondly the fast moving parts of your video seem to lack motion blur which says to me your shutter speed is wrong. You can switch the camera into shutter priority mode and manually set your shutter speed, for 24p it should be 1/48. One of the problems with this camera is if you select manual exposure control it will change shutter speed to 1/60 which is a bit of a minor PIA. Probably the difference between 1/48 and 1/60 is not going to be huge when it comes to motion rendition but why, oh why doesn't Sony let you get this right :(

Bob.
johnmeyer wrote on 1/23/2012, 4:07 PM
I downloaded the full, unaltered clip from YouTube. Of course YouTube does things to the clip, but FWIW, the downloaded version reports 23.976 and not 24 fps.

I then applied color curves and got this result (single frame snapshot):



Here's what the original frame looked like:



I did this because, to me, the results from the plugin posted earlier seemed to gain the highlights way too much, making the wall in the background appear too bright.

Here is a snapshot of the color curves settings I used:

amendegw wrote on 1/23/2012, 4:16 PM
"Secondly the fast moving parts of your video seem to lack motion blur which says to me your shutter speed is wrong"Bob, you're seeing the same thing as I. It's really counterintuitive. You'd think that you'd want a faster shutter speed to capture fast motion, but that only makes the video look staccato. The motion blur of a slower shutter speed helps smooth the motion significantly.

Which brings me to another point - Sony Vegas allows you to add "motion blur", but every time I've attempted to add it, it has made things worse not better. Is there a secret to using "motion blur" in post?

...Jerry

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

amendegw wrote on 1/23/2012, 4:23 PM
"I did this because, to me, the results from the plugin posted earlier seemed to gain the highlights way too much, making the wall in the background appear too bright."Yeah, the FBmn has a Highlight Correction which is supposed to correct for this. I maxed out this correction. Frederic, if you're listening - it would really be nice to be able to add more correction here.

That said, I added the Color Curves after the FBmn Exposure FX, and if I'd spent more than 5 minutes on this, I'm sure I could have done better.

...Jerry

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

johnmeyer wrote on 1/23/2012, 4:39 PM
Jerry,

I wasn't trying to be, in any way, critical of what you'd done. I assumed, since you know more about this than I ever will, that it was exactly as you said above, some sort of setting in the plugin that was being too aggressive.

The good thing about a dedicated plugin, is that it should give you good results more quickly and it may provide more automatic corrections if the lighting changes. The good thing about color curves is that it is free, and you can get very exact control over exactly how much each part of the histogram gets altered.
amendegw wrote on 1/23/2012, 4:47 PM
John, I think I came off as being more insecure than I really am. What I love about this forum is that people attack problems from different directions and come up with some really good answers.

In the meantime, I, for one, am learning a lot from others techniques.

...Jerry

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9

farss wrote on 1/23/2012, 5:06 PM
"Which brings me to another point - Sony Vegas allows you to add "motion blur", but every time I've attempted to add it, it has made things worse not better. Is there a secret to using "motion blur" in post?"

The "secret" is it is almost impossible to add motion blur in post that is anything like what happens in a camera. Even the most expensive high end tools are not very good at this.
Think about what happens in a camera. As the image moves over the sensor the photons accumulate, be it digital or film. This yields a smooth blur.
To do this in post you would have to be able to compute where every pixel moves between pairs of frames and add blur along the motion vector, no trivial task.
What Vegas and AE do is to use adjacent frame and blend them into the current frame. It can sort of help and Vegas gives you control over how this is done with Asymetric Gaussian probably the best. Still if you look closely you can very easily see it's just a bunch of unblurred frames composited together and may even make things look worse.

If you're working with something you're animating in Vegas or AE then you can do better by tweening the motion to get more "frames" to composite together, that's what Supersampling does. AE hides this more than Vegas does. Either way the results are much the same and still nothing like what happens in a real camera.

Twixtor does it better but you need to supply the motion vectors and some CGI apps can pass the motion vectors as metadata buried in the alpha channel I think so compositing apps running Twixtor can add somewhat better motion blur. None of this helps at all if it's something you've shot. Pretty much get it right in camera. This really is something you cannot fix in post.

Bob.
jerald wrote on 1/24/2012, 7:24 AM
This video footage is rather special in 2 ways: first, it is very consistent in content and exposure throughout, and second, the areas that need fixing (dancers' faces) have levels that are different than the rest of the scene (mostly). This allows some rather "surgical" tweaking of the color curves plug-in without tedious keyframes. Since you (Siby) mentioned that the *dancers faces* (sic) are dark, I purposed to address this, specifically.

The scene (to my amateur eyes) looks well exposed in general, except for the dancers, so I attempted to 'zero in' on the levels that needed fixing. To do that, I decided to find the part of the color curves (i.e. the portion of the diagonal line) that is specific to the faces.

To do this, I created a narrow, vertical 'spike' on the curve by creating a cluster of 3 anchor points on the default curve, then dragging the middle one up. Then, with the 3 points that make up the 'spike' selected (via lasso), I moved the 'spike' on the curve left & right to find the max & min levels that occur on the faces.

Then, on the 'default' color curve, I put anchor points at the min & max levels of the faces to create a color-curve-segment that applied primarily to the faces. I right-clicked these two 'boundary' anchor points and de-selected 'lock tangents.' With these steps completed, I am able to adjust the parts of the curves that affect the faces without changing overall exposure of the scene.

By dragging the tangent lines within the segment that affect the 'faces' segment of the curves, I was able to increase levels of the faces without changing overall exposure of the scene. Again, fortunately, this approach works with this particular footage because of its consistency as well as the natural relative isolation of the range of levels that needed to be fixed.

I hope this makes sense & is helpful.

Jerald
Siby wrote on 1/28/2012, 8:41 AM
Hi Jerald, Could you share the test veg file with me.
Frederic Baumann wrote on 1/29/2012, 4:53 AM
Hi all,

Thanks Jerry for the feedback regarding boosting highlights correction. I am currently working on a new version of Exposure, and will certainly consider this remark. This new version would include an additional slider to restore the saturation of corrected colors (boosting darks usually destroys saturation, but I found a mean to have it kept to some good extent, and this could be controlled through a slider).

Below are 4 stills:

1) the original picture (got from the Youtube video mentioned in the first post of this thread)
2) correction with Exposure GPU (highlights being dimmed)
3) a preview of the correction with saturation adjustment
4) a preview with more highlights control (value of 1.5 instead of 1)

By the way, I would be glad to apply these settings to the full video, but could not find how to download it... Please let me know (Siby?) if you are interested.

Any feedback welcome (Jerry and others!),
Frederic - FBmn Software

Original:


My Exposure GPU settings:


Preview of saturation restoration:


Preview of saturation restoration + advanced highlight control:

amendegw wrote on 1/29/2012, 7:23 AM
Frederic sent me a personal email on this - that's great customer service! My opinion - these improvements are good ones.

...Jerry

System Model: Alienware Area-51m R2
System: Windows 11 Home
Processor: Intel(R) Core(TM) i7-10700K CPU @ 3.80GHz, 3792 Mhz, 8 Core(s), 16 Logical Processor(s)
Installed Memory: 64.0 GB
Display Adapter: NVIDIA GeForce RTX 2070 Super (8GB), Nvidia Studio Driver 527.56 Dec 2022)
Overclock Off

Display: 1920x1080 144 hertz
Storage (12TB Total):
OS Drive: PM981a NVMe SAMSUNG 2048GB
Data Drive1: Samsung SSD 970 EVO Plus 2TB
Data Drive2: Samsung SSD 870 QVO 8TB

USB: Thunderbolt 3 (USB Type-C) port Supports USB 3.2 Gen 2, DisplayPort 1.2, Thunderbolt 3

Cameras:
Canon R5
Canon R3
Sony A9