Look: In-Camera or Post-Production?

Thanks! Actually, all of that makes total sense to me... BUT... does any of this stuff truly effect a finished image... I mean, YES, it does... but is it discernable?
 
Depends on the codec you're rendering to, and how many times you re-render.

In the old days we used to have to render out and "flatten" our effects and then re-import that footage every once in a while. Nowadays it's not so bad. If you don't render out your footage, it won't have any effect.

If your workflow is: put original footage on the timeline, apply filters, and render out to DVD -- then you won't be losing anything.
 
Barry_Green said:
If your workflow is: put original footage on the timeline, apply filters, and render out to DVD -- then you won't be losing anything.
Click to expand...

All right, that was exactly what I was wondering. This is my workflow when color grading, as I do everything within my NLE (Avid).
 
And you NEVER render until you go out to DVD? That is, until you are completely done with your project?
 
Brandon Rice said:
All right, that was exactly what I was wondering. This is my workflow when color grading, as I do everything within my NLE (Avid).
Click to expand...

That is fine for color grading, but keep in mind that if you don't capture to the full dynamic range (flat lighting) you can still take a pretty big resolution hit (see my previous explanation) when you modify (expand) the white and black points in your final render.

This is really easy to see, just shoot a gray scale or chip chart and look at the intermediate steps to see the loss in resolution. No magic here, its just the nature of any sampled system (pictures, films, music, etc).
 
Blaine said:
And you NEVER render until you go out to DVD? That is, until you are completely done with your project?
Click to expand...

That's true for me about 50% of the time, though I'll usually render a low-res proxy for sound mixing.
 
Well... I always keep my Video Filters seperate of my raw video files. It works a little different in Avid. I actually never have to render while in the NLE.

khmuse said:
That is fine for color grading, but keep in mind that if you don't capture to the full dynamic range (flat lighting) you can still take a pretty big resolution hit (see my previous explanation) when you modify (expand) the white and black points in your final render.

This is really easy to see, just shoot a gray scale or chip chart and look at the intermediate steps to see the loss in resolution. No magic here, its just the nature of any sampled system (pictures, films, music, etc).
Click to expand...

Your point is well taken... when I say flat, I don't mean flat in terms of dynamics, I mean flat in terms of keeping the black levels higher...
 
Brandon Rice said:
Your point is well taken... when I say flat, I don't mean flat in terms of dynamics, I mean flat in terms of keeping the black levels higher...
Click to expand...

This is the same thing. If your black levels don't extend to the bottom of your dynamic range, you simply start acquiring information (samples) at a level that is not zero. So, you still have reduced the dynamic range. I don't mean that you should crush the blacks to levels that they can't be recovered from, but that you should consider controlling the shadow detail (by lighting control) to a point that puts your darkest blacks at the bottom of your acquisition systems capability. The same is true for your highlights, light (or control them) so that they just reach the top of your systems capabilities.

Since you seem to be a Photoshop fan (me too), let me give you a way that you can see (and perhaps better grasp) these concepts. Take any scene and grab a frame form it and then import it into Photoshop. Use the histogram tool and look to see what your dynamic range is. If you have shot too flat (blacks too high in illumination level) you will have reduced your dynamic range and have not used the resolution that the system is capable of.

Sure you can push it around in post, but if you shot with 6 bit levels, then you will never have more than 6 (64 levels) bits of resolution. You can rescale anything into any new space and hit the top and bottom (you can go to 16 bit and have a 65k to 1 range) but if you shot 6 bit, you still have only 64 levels in between the expanded upper and lower range.
 
That all makes sense, and agree with having the blacks black, but not black beyond recognition (except some cases you might want that)... but again, I go back to saying, we can talk numbers all day long, but on screen will you see a noticeable difference?
 
I can easily see the difference, but this might not be important to everyone. I love numbers (all engineers do) so I can't help but use mathematics in everything I do (I know, I am more than a bit of a dweeb, but I have learned to live with it).

If you get it really close when shooting, and your post work flow is optimized in terms of preserving the resolution, you too should have no trouble seeing the difference that getting it right in camera can make.

Give it a try and see if you like the results. I suspect that you just might.
 
I plan on doing some tests in pre-production on a feature this summer, with the HVX, it includes some lighting tests in specific regards to this very topic. I am curious to see it side by side for myself.
 
Explorative Topic, Brandon.

I have a question for those who know a lot about compression, directly related to a "post" workflow: If you were to shoot "flat" with an intent on going with a hyper-stylized look, by exporting your entire film (or a segment) into a PNG format, batch-applying in Photoshop, and re-importing the saved PNGs into an NLE... have you undergone any compression? More detailed breakdown of the question--

A) Export from NLE in PNG frames.

B) Correct one image, apply to every image in the target sequence and save as PNG.

C) Import back into NLE as PNG on the timeline.


Have you recompressed or anything?
 
It all depends upon what you do to it while in Photoshop. The same dynamic range rules apply to all post work flows. You will never achieve more resolution than how it was shot. If you don't use the full dynamic range of the acquisition system, you limit what is possible.

For the life of me, I can't understand why anyone would want to limit their final product by intentionally shooting a flatter image than they want for the final output? If you had a high resolution system (say 12 bit or more) then perhaps, but at 8 bits it seems rather silly to me to give away a quantity that you have so little of to begin with (8 bit).

I really wonder what makes this approach attractive to some?
 
khmuse said:
It all depends upon what you do to it while in Photoshop. The same dynamic range rules apply to all post work flows. You will never achieve more resolution than how it was shot. If you don't use the full dynamic range of the acquisition system, you limit what is possible.

For the life of me, I can't understand why anyone would want to limit their final product by intentionally shooting a flatter image than they want for the final output? If you had a high resolution system (say 12 bit or more) then perhaps, but at 8 bits it seems rather silly to me to give away a quantity that you have so little of to begin with (8 bit).

I really wonder what makes this approach attractive to some?
Click to expand...

If you were to only "Grade" or "color correct" in photoshop. Not looking to uprez the resolution or sharpen or anything like that. Merely change this color to that or apply certain effects without suffering video degradation. Is that possible?

I'd like to say that shooting "flat" or whatever could be useful for Special Effects shots or plates. The only other reason I've come to understand is that shooting flat and crushing in post and all that lets people eliminate noise?

I thought it was because they aren't establishing a look for the piece before going into production, and want to figure that out later. Difference is, with pro-productions, there's a look established and it's straight-forward. "We need it to look like this.". I suppose with post-workflow options available, you can make that decision later?
 
On my histogram in PS... if there is a ! symbol, does that mean the image has things outside the dynamic range? Just curious.

Also, I personally like shooting this way as I like being free to change the look of a scene how I see fit. I usually never stick with the initial look I give a scene. There is a lot of tweaking involved.
 
Kholi said:
If you were to only "Grade" or "color correct" in photoshop. Not looking to uprez the resolution or sharpen or anything like that. Merely change this color to that or apply certain effects without suffering video degradation. Is that possible?
Click to expand...

Certainly rescaling the image size is going to have a qualitative impact. Everything you do, will have some consequences, but these may be outweighed by the final results.

Kholi said:
I'd like to say that shooting "flat" or whatever could be useful for Special Effects shots or plates. The only other reason I've come to understand is that shooting flat and crushing in post and all that lets people eliminate noise?
Click to expand...

This isn't a great approach. You can achieve the same by illuminating the scene to a level that gets the camera out of the noise floor and using the master pedestal to set the amount of shadow detail you want. This makes better use of the dynamic range of the system. Remember that the sensor (CCDs) are analog devices but the quantizer is limited by it's design (8 bits in these instances).

Kholi said:
I thought it was because they aren't establishing a look for the piece before going into production, and want to figure that out later. Difference is, with pro-productions, there's a look established and it's straight-forward. "We need it to look like this.". I suppose with post-workflow options available, you can make that decision later?
Click to expand...

I try and determine what the goal is in the pre pro stage and then do my testing prior to the beginning of principal photography, then all I have to do is realize the goals (as best I can with the resources available).

I guess I am not a gorilla approach type, but try to bring the same values to a video project as I would a film one.
 
khmuse said:
Certainly rescaling the image size is going to have a qualitative impact. E
Click to expand...

This is a given, and should be for anyone who understands the basics of photo correction and how it works. I would have no intent on upscaling or anything like that.


This isn't a great approach. You can achieve the same by illuminating the scene to a level that gets the camera out of the noise floor and using the master pedestal to set the amount of shadow detail you want. This makes better use of the dynamic range of the system. Remember that the sensor (CCDs) are analog devices but the quantizer is limited by it's design (8 bits in these instances).
Click to expand...

I haven't actually tried any sort of "do-it-in-post" method, because I haven't had anything worth the effort. However, I think I will try it this week by doing another "household chores.mov" and doing all sorts of odd things in FCP, Motion, and After Effects to see what comes out of it.

If you were trying to do what KHMUSE has just described, what would the settings be equivalent to on an HVX?

I try and determine what the goal is in the pre pro stage and then do my testing prior to the beginning of principal photography, then all I have to do is realize the goals (as best I can with the resources available).

I guess I am not a gorilla approach type, but try to bring the same values to a video project as I would a film one.
Click to expand...

I work around a lot of people who've created by that approach, the do-it-in-camera. They all suggest that doing it in post is NOT the way. Don't put off tomorrow what you can do today.

Hopefully when I do something worthwhile I'll have a real opinion on post versus in-camera.
 
Also, food for thought . . .

If you don't have a pretty detailed vision for the final product when you go into production (of which the final color grading is only a component), you're far more likely to waste a lot more time during production, and also ultimately end up with things you wish you had done differently or have to go back and redo.

Not saying that doesn't happen with well-planned shoots -- it does -- but the risk of it is much, much lower, and the post-production process is much smoother.
 
Anytime you manipulate an 8-bit image, you effect the quality. There are a finite amount of steps/gradations to work with. Particularly when adding contrast in 8-bit images, you are stretching out the midtones which can cause banding.

Magic Bullet does a little better in this regard because it converts the image to 16-bit float per channel and then back to 8-bit. It's a little smoother than what you get out of standard NLE adjustments, but it still has it's limits.

If any adjustment can be harmful, then why make any adjustments? Well, the advantage of tweaking each shot so that it matches perfectly in color cast/contrast to the shots before and after it is beneficial... perhaps more beneficial than that technically negative effect of manipulating the image.

Master pedestal is an interesting case, because there is an emperical method to determine the correct setting... and someone has already gone to the trouble. With the lens cap on, ND 1/64 engaged, and iris CLOSED, you would think that you would get a digital value of 0 if you recorded a few seconds of video, right? Well, it depends on the master pedestal setting. I don't have the link handy, but someone did this analysis on the DVX, with every MP setting. They found that -6 is the master pedestal setting that results in a digital 0 under those conditions. -5 resulted in blacks that were not true black, and -7 lost detail. I don't remember what the gamma/matrix settings for their test was, and that is an important control.

Does shooting with a higher than -6 MP record any additional detail? I don't believe so. Does it shift the shadow detail into a range where there are more subtle tonal variations? I'm not sure. However, particularly with video, you need to maximize the dynamic range, which means protecting the highlights and using an MP that gets you true blacks without crushing. Ideally there would be a histogram on the LCD to allow you to monitor this real time.
 
You must log in or register to reply here.
Back
Top