Alright guys, again I'm going to ask for your help. I'm trying to fake 35mm depth of field in post. I started with the easiest clip (no idea how I'm going to do it to all of them yet, but I'm trying to see if this is possible.

Now, I've stripped away ALL color correction so this is raw footage, done so that you can see the DOF effect better. The compression (Sorensen 3) affects it somewhat but not so much that you cant see the effect and, I hope, judge it for realism. The clip is 3.5 meg so it shouldn't be too much of a strain.

http://www.birthofthecool.com/Testing/dof_test.mov

Please let me know what you think, and if it sucks, why so.

Matt

Well since you mentioned it I looked for the fake DOF not working properly and know what to look for. If you look at the cabinet right by the edge of his face you'll see it's not blurred, which is what happens when you try to mask things.

If I were watching it without knowing fake DOF was applied, I probably wouldn't notice it.

Alternate ways to draw attention to the guys face would be:
A- Add vignetting... i.e. a mild black gradient from left to right. The black matte generator and the masking tools in Vegas is one way to do it.

B- Make the blurring drop off as it gets nearer the guy's face. That isn't quite the same though.

C- If the colors are different enough, you can use secondary color correction to punch away everything except for the guy's face. I'm not sure if it can be done in this scene, you have to very carefully limit things by color and luminance. Brown is a dark shade of orange/flesh tones.

You told me about it and I knew where to look, as Glenn also mentioned.

Otherwise, I would have not know! Great job, I'd say! :)

It works for very specific situations. If you have a pan or tracking shot, that would make things very hard.

But I doubt that in the middle of a movie many people would notice it was done in post, so for this case works quite well.

Well, It's getting better. Right now I'm experimenting with using a b-spline mask in Motion and keyframing the mask to move with the subject. Pans without people movement are actually very easy to do like this. It's people movement that is difficult, but I'm getting MUCH better results this way. Will post a new clip soon - well as soon as I can figure it all out.

Matt

great effect matt, whats the basic proceedure for doing this...???

It did look good!

That shot looked like it would have been easier to pull off an in camera shallow DOF. But the option to do it in post looks good.

Thanks guys, I'm shooting, arranging locations and doing some basic editing at night (all while holding down my day gig) so I forgot to respond.

The basic premise is to create shallow DOF in post that resembles that of film (assuming you couldn't/didn't get it in camera). So I started the experiment (that you saw) by bringing the sequence into Motion (or AE). I had taken a frame grab from that sequence and worked with it in photoshop to extend the hutch and get rid of the actor. Then I blurred the image. The problem was to make the focus fall off in a realistic manner. So I first drew a line of perspective for the shot, this was basically a line from the lower right hand corner to the upper left hand corner. I then cut the image up into diagonally striped sections moving along that line. I then applied a gaussian blur to each section, making it more blurred for each further section. I feathered these sections together so that the effect was seamless and that was it. I then brought that into motion and layered the rough mask of the actor over it.

What I'm doing now is similiar, except instead of using a still (different shot) I'm creating bezier masks for the background and applying affects to the masks that I then animate to move with the shot - in this case a dolly tracking shot. We'll see, so far it's not to bad, but it still needs some tweaking.

Matt

I tried a small app once, it was freeware. What it did: take to layers of video - one with the background alone, the second with the actor on the background. Then, it would substract the first video from the second to leave the actor without background. Kind of green screen but without the green screen! :)
Now you have a video with the background (that you could blur or whatever...) and another video with the actor alone.

Of course, the scene should be static and nothing should move in the background...

ost some more videos but with more action. for instance 5 character walking to a school or something. A moving shot would be cool as well. Lots of detail in the background.

nice video

chich cam did you use>?|

It's late and i'm bored, so i captured a couple of screenshots from a promo piece i'm working on. The darker more 'contrasty' areas are after fake-dof and basic color correction (final color grading will be applied after the piece is approved)..

For those at work, i should note, the 2nd image has a girl in her underwear/bikini.

http://www.adamgeek.com/dvinfo_dof1.jpg
http://www.adamgeek.com/dvinfo_dof2.jpg



<<<-- Originally posted by Mike Hardcastle : great effect matt, whats the basic proceedure for doing this...??? -->>>

the procedure is basically rotoscoping.. you can search google for many tutorials. i'm not sure what he is using to do this in, but after effects / combustion would be the usual choice.

<<<-- Originally posted by Cosmin Rotaru : I tried a small app once, it was freeware. What it did: take to layers of video - one with the background alone, the second with the actor on the background. Then, it would substract the first video from the second to leave the actor without background. Kind of green screen but without the green screen! :)
Now you have a video with the background (that you could blur or whatever...) and another video with the actor alone.

Of course, the scene should be static and nothing should move in the background... -->>>


You wouldn't happen to remember the name of this program would you?

I could use this in a project I'm currently working on.

Thanks.

I'll search it when I get home. I don't even remeber its name...
I'll let you know tomorow.

Please, please, if you could find this program it would make my life SOOO much easier!!!!

<<<-- Originally posted by Matthew Cherry : Please, please, if you could find this program it would make my life SOOO much easier!!!! -->>>

if your goal is simply to blur the background on a fixed shot where you can control the background behind the subject.. rather than find a plugin or program to do this, why not just shoot the interview in front of greenscreen anyway. a piece of 3'x3' posterboard would work fine, since you really only have to cover the area of the frame around the subject's head/torso. then shoot a 2nd static shot lit the same without the subject in it, chromakey out the posterboard, and composite the matte over the 2nd static shot. with pratice i'm sure you could produce good results. 'course.. if you've already shot the pieces you need and you can't reshoot them, my suggestions don't help much.

I couldn't find the software... I'll try to google it, if I could find the right words...

Have you seen this (althought it is not done in post):
http://www.indietoolbox.com/

(it has been posted here before, and someone sugested silk instead of.. that)

here is something some may find interesting:
http://borg.cc.gatech.edu/Software/bsubtract/
It seems so complicated to use (for me) that I can't follow the entire explanation... although, the pics at the end are very suggestive!

Perhaps?...

1. Make a "depth map" with varying shades for background objects depending on how far away they are supposed to be (this is the hard-work part).

2. Apply a Z-Blur or lens blur filter (with true lens blur characteristics) to each depth/layer (you can also set this over time using keyframes for rack-focus) using filters designed for this:

Image Lounge (http://www.redgiantsoftware.com/imagelounge.html)

CinemaFocus / IrisFilter5.6 / FinalFocus2.0 (http://www.reiji.net/op-e/index.html)

Lenscare
(http://www.frischluft.com/lenscare/introduction.php)

BCC Z-Blur Filter
(Boris Continuum Complete - http://www.borisfx.com/download_files/bccfilters/BCC_Z-Blur.pdf)

Defocus and Seelctive Soft Focus
(Digital Film Tools 55mm and Composite Suite)

S_ZDepthCueBlur, S_RackDfComp, S_RackDefocus
(Genarts Sapphire)

The Depth of Field effect
(After Effects 6.5 Pro)

There are others...


(Matthew, I was actually surprised that you are using a gaussian blur; I gathered from other posts that you have 55mm and Composite Suite)

John



Another method tutorial here (with sample movie):

http://fxhome.com/support/tutorials/view.php?i=8

sample movie: http://tarn.fxhome.com/tutorials/fakedof.mov

John, great post, will continue to experiment with what you are saying, although it may take a bit to digest what you have just spelled out.

Yes, I have all of DFT products, but I only have them for Final Cut Pro and After Effects (although I have used the AE plugs for Apple's Motion. I don't have the plugs for photoshop. and I did this effect in photoshop. Also I just purchased them and am really just learning how to use all of them now.

I don't know how to make a "depth map". I mean I think that's what I did in photoshop, by slicing the image into layers and then applying a varying amount of blue the further away you went.

If you would be so kind, I think there are more than a few here who are interested, could you explain what you're suggesting in a bit more detail when you get a chance?

Best,

Matt

My initial thought was:

1. Bring the footage into After Effects
2. Create a composition based on the footage
3. Create a solid (with black as the color)
4. Set the opacity of the solid to zero, mode to darken
5. Create masks on the solid for the different "depths" of the footage
6. Keyframe the masks
7. Set the solid opacity back to 100%
8. Change the opacities of each mask based on the depth
9. Output the solid as a "depth" mask (with white beneath the transparent areas which are meant to be the background, or farthest away)

(This was the easiest way I could think of to create an accurate depth mask)


But, after some testing Sunday night, I think that in most cases it is easier to rotoscope the footage (unless you need to do something based on depth in other applications - then a depth mask is probably best). I did the following with pretty good results:

1. Bring the footage into After Effects.

2. Create a composition from the footage.

*Deinterlaced footage is best (and besides, this is the "Towards a Film Look Using DV" forum). Higher resolution is best. And obviously, footage with minimal motion would be best --- having the camera locked down (tripod) will make things easier. And obviously, long clips will take a lot of work --- probably best for short clips.

3. Determine the "depths" for objects (including people) or areas... I just drew an overhead map on some paper showing what needed to be in what focal position.

4. Create duplicates of the footage on the timeline --- one for the background and one for each "depth".

5. Scrub the footage and locate the extreme points of motion. Place a marker on the timeline for each of these points (for each layer).

6. Using the mask (pen) tool, create masks for each set of objects in each layer. For now, change each mask's mode to "none" so that you can see the entire frame while working.

***It is easier to create multiple overlapping masks for a single object if that object will be moving and changing shape (for example, a turning head may cause the nose to disappear and reappear--- so I would create a head mask and a nose mask on the same layer). Using multiple masks even for a single object (like the head) allows you to apply different characteristics, such as feathering, to one edge of an object so that you can realistically blur the far side of a face, for instance (You can also apply motion blur individually to a mask if necessary). --- Separate masks also have the benefit of faster rendering, because certain effects, such as motion blur, only apply to them, not to all masks or a larger area.

7. When all the masks are created, begin adjusting / keyframing the mask (or set of masks) one at a time. This will be the most time consuming part. Complex objects (people) will require finer point adjustments while simpler objects (such as a book) may just require moving the mask. To be safe - Select the masks and right click - "Hide locked masks", then open the layer properties on the timeline and turn on the lock for all of them. Then just unlock the one (or group, such as head and nose) being adjusted.

8. Begin with the first mask by selecting it in the timeline (be sure timeline is at the right point), adjusting the mask shape (point by point) and then clicking the stopwatch to create a keyframe. Now move to the next marker (the extremes you marked) and adjust the mask again for another keyframe. After Effects should interpolate in between pretty well (though you have to go back and adjust these later). Work through all the markers for that set of masks, then scrub and repeat for the in-between frames, until you are satisfied with the result. Now lock the mask and unlock the next one on the next layer for adjustment.

9. When all masks are adjusted and keyframed to your liking, unlock all the masks, and change their modes back to “Add”.

10. Now apply lens blur (with a lens blur filter that provides accurate characteristics) in different strengths to each layer as needed.

11. Adjust the feather (and other properties) for each mask as needed.

12. Render your beautiful rotoscoped footage with Depth of Field. :-)

Matthew,

I've gathered better information over the last couple of days and will be posting a much more accurate and detailed post soon on depth masks and dof (after a few tests).

John

Excellent. Hey, do me a favor. Go to the filmlook section and check out the grabs I posted. I would be interested in your opinion if you have a moment.

Thanks

Matt

Im confused....sort of!! I didn't think depth of field had so much to do with film or the medium, as it did the lenses. Can anyone help?

It isn't that Depth of Field (DOF) is something that is a characteristic of film; it is that the abilty to truly control DOF is something that is seen as an intrinsic part of professional movies --- the abilty to control DOF (specifically shallow DOF) is thus seen by many as part of the "film look". Most consumer (and prosumer) camcorders lack sufficient long-focal-length lenses to achieve adequate shallow DOF (in my opinion). A movie doesn't need shallow DOF --- but it is a tool that many of us often lack. Besides, the ability to add the shallow DOF effect (realistically) in post can be useful if the scene was shot without it.

John

nice try!! it looks all right! I think the background needs a little more softening/more blurr. FOr the rest, nice work!