Creating Post Production Depth of Field

[Walter Graff]

I mentioned a technique I used to create depth of field in post in another thread here and some folks asked to see it.

 

Post has become key for me as lf late with many of the low to medium size budget spots I create. Many times I am unsure of the way I want to go in post, and/or don't have all the great tools in the field to use. And sometimes I just don't have time nor crew in the field so know that post can be a great extention of what I do in production. For example when it comes to 24p effects I have found shooting everything in 30 and using Natress 24p filters in post is a a great solution when I am unsure of how I want to go in post look-wise. Graham has one of hte best 24p plug ins out there as far as I am concerned and it's become as importat a part of my production as anything else. He is part of the RED team software wise now and has always been on the forefront of technology. His filters allows me to create exactly the same in-camera 24p pull down effect in post but with far more control. I say control as in times when a pan is too much or action too much or something a bit off for regular in-camera 24p, I can make minor tweaks using Natress controls in post to create smoother looking motion, etc.

 

I use the gamut of cameras from my JVC HD200 to HDCAM on a weekly basis. I shoot pretty much all of my local TV spots on my HD200 with a DTE hard disk recorder in HD mode even though no one accepts HD for spots. I love the combination of disk recorder and the camera. The hard drive method beats tape and solid state by a mile becaue I have hours of recording so I dont need to worry about running out of space. And I throw in a tape and have instant archiving so never need to worry about data I might delete and need again later. It allows me to edit immediately by simply unplugging the HDR from the camera and into my edit system with no transcoding, etc. Of course with 1/3" cameras I don't always get the DoF I would like so I have been using what I call the "Graffield filter" to create depth of field in post. I love the results. It allows me to add as much softness to a shot as I want and the results are excellent

 

Below is a still showing raw footage uncorrected and then Graffield depth of field applied with color correction. And there is a link to the actual spot below those so you can see it in action.

 

 

http://www.film-and-video.com/mazda2.mov

[Chance Shirley]

Very interesting.

 

Did you explain how the Graffield works in the other post? If so, could you provide a link (couldn't find it via site search)?

 

Thanks...

[Alessandro Machi]

Is solid state still prohibitive cost wise? The stuff has been out for a few years now. The combination of hard drive and tape seems like a great combination.

[Walter Graff]

Is solid state still prohibitive cost wise? The stuff has been out for a few years now. The combination of hard drive and tape seems like a great combination.

 

Solid state is getting better but still not as easy, and far from cheap. I can record seven hours of HD on my hard drive, plug it in and edit, no waiting. And having the tape is the only way to archive successfully. And best of all, I do it as I record so don't have to spend hours later laying stuff off. Like many other myths that are all over the web, someone said that hard disk recording was hard, or difficult and most folks belived it even though the person that said it had never even seen a picture of a hard disk recorder or the company that said it was trying to push their expensive solid state recorder that is like trying to light your dinner by candlelight using only matches one at a time.

[Guest tylerhawes]

So are you drawing a mask around her with feathered edges and applying the filter outside the mask, or what? How is it different from a blur filter?

[Tim O'Connor]

Walter, from many of your posts I have the impression that you're not crazy about the HVX-200. Since it's in

about the same price ballpark as the JVC GY-200, is the JVC a camera that you like a great deal more and if so

can you say why or perhaps is it that you like it for your purposes as much as any camera in that price range?

 

I'm not challenging the JVC, I haven't used it but I do have an HVX-200 and for my purposes, which do

not include lots of tv spots, it works pretty well.

 

Also, c'mon, how you'd you achieve the Graffield effect? Pretty cool by the way.

 

What brand DTE hard drive recorder do you use? We just bought three Firestores for editing multiple camera shoots in

Multi-Cam in Final Cut Pro and they haven't been the vastly more efficient solution that we had thought. I haven't been the one doing the editing for those programs but evidently even though the capture time is reduced, there still is some signifiicant time required to get the footage into FCP. It's not the instant drag and drop situation that we had understood it to be.

 

By the way, the audio in that spot is great throughout and really consistent. When she's sitting in the car in the showroom it's striking how clean and clear it sounds.

[Walter Graff]

"is the JVC a camera that you like a great deal more and if so

can you say why or perhaps is it that you like it for your purposes as much as any camera in that price range?"

 

In my tests of all the cameras in this price range, I fould the JVC created the best looking picture (subjective, and objective quality-wise when using a chart to measure how well it produced an HD picture), has the best ergonomics for a pro camera, and offers me the ability at using a pl mount lens adapter which makes unbelievable pictures. The HVX200 scored very poorly for actual picture quality when compared to the other cameras in the same price range in my tests. It was a great marketing ploy for Panasonic who simply took a DVX100 added a bit of circuitry making HD through pixel shift rather than through actual HD sensors, added P2 which was a method of recording they tried in the pro news world but failed to catch on so they made their money back making into a prosumer method of recording with the HVX. By itself it looks great but once you see it side by side as I have often you can't see it as making a very good picture. My tests involve a month of extensive shooting with all. Here was my initial tests of the HVX and HD100:

 

http://www.bluesky-web.com/HDVHVX.htm

 

What I found with the HVX with my tests was a camera that simply did not do what is advertized when compared to cameras in the same price range. I am sure it works well for you. I have too used the HVX200 for jobs and it worked well too. But for my tastes P2 is not my cup of tea and I found that the other three cameras in the price range offered far more picture for the money. And now with Sony's latest entry, I think the bar is raised even more.

 

Of the four, Canon made a fantastic picture but ergonomically is a piece of crap to me. Sony was also very good, but I needed a more shoulder mount camera. So the JVC was my choice.

 

"We just bought three Firestores for editing multiple camera shoots in

Multi-Cam in Final Cut Pro and they haven't been the vastly more efficient solution that we had thought."

 

Being able to record hours of HD rather than having to pull and transcoding cards every 40 minutes is far easier for me. Plugging a drive in, dumping all footage to a house drive and instantly editing works great for me. You must be doing something differently than I.

 

 

By the way, the audio in that spot is great throughout and really consistent. When she's sitting in the car in the showroom it's striking how clean and clear it sounds.

Good audio sure makes a spot look better. I sweeten all my spots. Having worked in the audio business when I started, I like clean audio. As for the effect, it's not so much that it is special, just that there are ways in post to create effects we are so used to in camera. Since I had no time or crew to deal with my lens adapter which would have given me shallow depth of field, and I do not use those spinning lens adapter things that degrade pictures overall, I went with my third option.

[Will Earl]

I hope your not using a Gaussian Blur? Because it isn't a very effective DOF effect - too smooth, with no depth or shape to it.

 

Try utilising a 'lens' blur instead (I'm not sure what tool your using to do this, but most NLE/Compositing/Photo apps these days include the filter). There are filters out there which produce much more realistic out-of-focus effects.

[Walter Graff]

Nothing beats true Dof in camera but as my demo shows, you can create very convincing looks in post when need be. I could spend hours playing with the filters I use to make changes in distance and blur, but for the purposes of this commerical, no one cares. It does what it needs to do.

[Timothy David Orme]

A still photo of this technique might look okay, if done properly, but in general I think it looks as fake as it is.

 

I have a friend who does this all the time: masks out the subject, then creates a blurred layer beneath it. Generally it looks really bizarre once it's set in motion (and yes, he does move the mask with the subject), almost dreamlike.

 

It's a good technique for what it is, but it doesn't look like 'natural' depth of field to me.

[Walter Graff]

A still photo of this technique might look okay, if done properly, but in general I think it looks as fake as it is.

 

I have a friend who does this all the time: masks out the subject, then creates a blurred layer beneath it. Generally it looks really bizarre once it's set in motion (and yes, he does move the mask with the subject), almost dreamlike.

 

It's a good technique for what it is, but it doesn't look like 'natural' depth of field to me.

 

 

How about when I soften the background on the first shot when I zoom in to her? Does that look better?

[Chad Stockfleth]

It's not always bad, but I tend to think it looks like a feathered key.

[Michael Nash]

The problem with artificially softened backgrounds is that it's easy to buzz the background so much that it becomes "optically" unrealistic in comparison to the foreground that's in focus, respective to distance and focal length.

 

If you think about focus as a cone that gets wider from the point of focus, softness progresses in a linear fashion. If you've got a person's head clearly in focus from front to back you wouldn't expect the near background to be extremely soft. Doing that creates the illusion that focus is behaving more like a trumpet bell instead of a cone, which ends up looking unnatural.

 

The technique is good, but in the talking head interview the background looks too soft to be realistic. I think taking it down a notch and it would be more believable. Also when you've got a lot of depth you might need to do multiple layers of defocus to create a convincing "cone of focus." You'd expect the cars to be sharper than the building across the street.

 

The zoom in/buzz was a nice touch though, and convincing.

[Will Earl]

Finally got a chance to look at the video, the use of it in the first shot looks better.

 

When doing effects like these, it's typically best to start with "what-actually-happens-using-real-lenses" and then start applying creative adjustments. Depth-of-Field is one of few characteristics of a camera that gives us a sense of scale, so messing with it too much can screw up an image. Even if the audience doesn't know technically why it's wrong, they'll still pick up that it's wrong.

[Walter Graff]

All nice advice, but cinematography advice. I'm not trying to make perfect DoF. I'm an advertiser trying to get my viewer to look where I want them to look. The commercial serves that purpose well. In fact it's a common technique used in both the lowest budget commercials and $250k spots I work on. It's a great effect for those that want to create varying degrees of softness to parts of a picture to draw your eye where we want it. Frankly if I showed the spot and simply asked folks if they liked it, none of you would have noticed what I did. I've already done that here on the site in the past with other examples and techniques in post and no one noticed. That is the problem with asking a loaded question, folks look for problems because you introduced thier subjective mind to want to find problems. One of the companies I work with spends millions of dollar a year on focus groups. I am still amazed after all these years on just how much effort goes into simply coming up with the questions asked so that they don't bias the product.

 

"the use of it in the first shot looks better."

 

"The zoom in/buzz was a nice touch though, and convincing."

 

As I said asking a question the way I did often gets folks to want to not like something or simply find the error because I am telling you I am doing something artificially. It's how magic tricks work. Your mind can't believe two things at once. A good trick allows you to think you are seeing something not real so you try to figure out the way it is done. Often the magician plays off that and makes you think you know. The second part of the trick defies you because as you are looking to find the error in the first part, a second part of the trick is introduced which confuses your already confused mind. Proof is that when I asked "How about when I soften the background on the first shot when I zoom in to her? Does that look better?" and got the results above I knew what that would elicite. I knew because I asked it to get you to say what I wanted. Fact is I did nothing to the first shot. It's natural depth of field just as it was shot. Just wanted to show how easy it is to convince someone of something. I told you I did it and no one knows I didn't. It works both ways. I was going to put up another spot where I did nothing but used in camera Dof the way it was and ask what you thought of the effect I 'created' there to prove the point but got lazy and didn't upload one. Had I done it, I would have gotten much of the same response because some of you now know the trick and want to find why it doesn't work for you whether I actually did it or not. But it's not that it doesn't work, it's that I let you in on the secret so as cinematographers you tried to rationalize it because I told you to in the way I asked.

 

Thanks for all the feedback!

[Paul Bruening]

Hey Walter,

 

I think you're doing interesting and nice work with this. I've noticed that, sometimes, folks react to a new idea in absolute terms, here. No idea is perfect. No technique is perfect. No technology is perfect. Given that, I think you've done a service to us by sharing this with us. Thanks.

[Michael Nash]

We didn't give you feedback. You just believed we have an opinion and know what we're talking about because we tricked you into thinking we do. Clearly none of us really have any clue.

[Michael McIntyre]

Nothing beats true Dof in camera... no one cares.

 

The unauthorized, late-night edit to your post. Come on, you rained on my DVX in-camera looks. I figure it's a fair trade. :P

[Damien Bhatti]

As I said asking a question the way I did often gets folks to want to not like something or simply find the error because I am telling you I am doing something artificially. It's how magic tricks work. Your mind can't believe two things at once. A good trick allows you to think you are seeing something not real so you try to figure out the way it is done. Often the magician plays off that and makes you think you know. The second part of the trick defies you because as you are looking to find the error in the first part, a second part of the trick is introduced which confuses your already confused mind. Proof is that when I asked "How about when I soften the background on the first shot when I zoom in to her? Does that look better?" and got the results above I knew what that would elicite. I knew because I asked it to get you to say what I wanted. Fact is I did nothing to the first shot. It's natural depth of field just as it was shot. Just wanted to show how easy it is to convince someone of something. I told you I did it and no one knows I didn't. It works both ways. I was going to put up another spot where I did nothing but used in camera Dof the way it was and ask what you thought of the effect I 'created' there to prove the point but got lazy and didn't upload one. Had I done it, I would have gotten much of the same response because some of you now know the trick and want to find why it doesn't work for you whether I actually did it or not. But it's not that it doesn't work, it's that I let you in on the secret so as cinematographers you tried to rationalize it because I told you to in the way I asked.

 

 

Sounds like a Neuro Linguistic Programming seminar!

[Michael McIntyre]

Sounds like a Neuro Linguistic Programming seminar!

Seriously. Yet another 'unauthorized' edit: "I didn't get DoF in-field. I blurred background in post. I'm a magician". :blink:

[Will Earl]

Proof is that when I asked "How about when I soften the background on the first shot when I zoom in to her? Does that look better?" and got the results above I knew what that would elicite. I knew because I asked it to get you to say what I wanted. Fact is I did nothing to the first shot. It's natural depth of field just as it was shot. Just wanted to show how easy it is to convince someone of something. I told you I did it and no one knows I didn't. It works both ways.

 

Walter, I'm sorry but I'm not entirely sure of what your point is or what it is your trying to prove here?

 

I get the feeling you started off this thread on a more serious-and-trying-to-provide-useful-info note, but have descended into a I-tricked-you-all-look-how-clever-I-am chord.

[Chad Stockfleth]

I think it looks like a bad key.

[John Sprung]

Below is a still showing raw footage uncorrected and then Graffield depth of field applied with color correction.

Looking at the two stills, I'm guessing that she's in front of a green screen, and you apply a blur to the BG element?

 

 

-- J.S.