Any Experts on DOF in Vue5I?

[Shaun_michael]

Hiya,


I'm no Vue expert, but I have a certain amount of photography knowledge. I've had a quick read of the Vue manual, and as soon as I see the words "increased render time" I glaze over, especially as Vue isn't exactly the nippiest of renderers.

Your desert photos will have a large depth of field, and thus little perceived depth, simply because you will have a lot of light, both direct and indirect bouncing into your camera lens (please excuse me if I'm teaching grandma to suck eggs here and telling you what you already know). Your camera will recommend a small aperture setting, and without getting into circles of confusion, will give you plenty of depth of field. Couple this with a wide angle focal length and you're looking at almost hyperfocal depth of field effects.

Considering the render times involved with vue, and the seemingly hit and miss method Vue of attaining DOF (not like lightwave where you can focus specifically on a null for instance) I would render without depth of field and do it in post using the z-buffer exported as you render in Vue.

At least this way you are sure to get your final render done with less effort and time, and then have full control over your DOF.

I use combustion for comping and have an excellent DOF plugin called lenscare (its available for After Effects too and compatible photoshop programs:-

http://www.frischluft.com/lenscare/introduction.php

It uses the z-depth you've generated to construct highly believable DOF effects, including the aperture element effects you see on out of focus highlights (you know, when you see a hexagon-shaped light when looking at car tail-lights when they're out of focus for example), and this is simulating the shape of your aperture as you're taking the shot - something that simple blurring methods does not do. You can change the number of sides your iris has. You can also adjust the z-depth clipping and point of focus, and it's pretty fast - certainly a great deal faster and controllable than rendering it all in Vue.

Anyhoo, I know this doesn't help you with learning Vue, and having read most, if not all your other threads I know this is your main objective, but I would still advise going this way in the long run for workflow purposes.

I've attached a couple of examples done very quickly and compressed a great deal.

Cheers,

Shaun

ps - Just spent the last 6 months training a TV studio dirctor all about what lenses do for him so I'm kinda in DOF mode right now. Sorry for being so long winded, and apologies as I expect you know all this anyway :D

EDIT: By the way, I forgot to mention that the two DOF shots were using the same z-buffer image just adjusted within the plug - all keyframeable.

[mdunakin]

No, Shaun!, that was all perfect! :)

I probably should of mentioned that I wouldn't mind doing any of this in post at all.
In fact, I'd prefer doing it in post, as I feel I would have more control, but also just the speed
difference alone makes all the difference in the world, and I spend enough time on renders as it is LOL

But, even with that, I should of said, that what I meant by doing the DOF in Vue,
I meant ANY method that uses Vue, which in this case would be using the Z Buffer
and how would I set that up or do I need to set up "anything" to make the Z Buffer work?
Ya know, does the Z Buffer just automatically make a depth thingy where
Post would know what it means and be able to deal with it?

I would more then likely be using PS to do the Post Work.
I have an older version of AE, and AE and me just never did get along.
My question from here would also be, can you keyframe a sequence with that plugin for PS?
So that I could run a sequence of images through the Actions and have that
action change over frames as they go along through PS?
Or can you only do this with something like AE?

I still haven't install a lot of my programs on this new system, so I do have some
cool programs, though some are older, like I think my AE is like v4.0 or something?

And I don't mind the baby talk when it comes to things like DOF, as I do understand how
it all works and pretty much about Z Buffers, but I'm just not that great at this area
of artwork in computers yet, as I rarely use DOF but am wanting to change my
wicked ways and START using more DOF, when it's appropriot :)

Thanx, and 89 bucks for the PS version is in my price range too!

.....................md :)
P.S. BTW, would I need to save out the images as something like psd or png
or something or do you need to save out the Z Buffer as some seperate thing?
Thanx!
EDIT: P.S.S. I recently upgraded to PS-CS2.
STILLLL haven't even looked through the extra CD with the new training tips and such!

.

[mdunakin]

Okay, from what I've been reading in the PDF, I see that the DOF is determined by
the focus point, which what I did, was to add several simple flat planes in the scene
(little squares) and set them to be Hidden from Render, and then just used those to set
the camera focal point for each camera, as this allowed me to place each Target Square
in a position that I wanted, without screwing with my camera's actual Target Point
that's built in and connceted to the camera by that little dotted line.
Now, after setting all of these, I did one test render with a far away scene,
just to make sure if this method worked or not, and yes indeed it did!
The Z Buffer, if you click on the last little icon on the little top right part of
a rendered frame, it shows the Z Buffer info as grey and you can easily
see the exact! DOF you set for it which can then be used to
adjust how much DOF (blur) you want later in post.

Works like a charm so far, now I just need to see
how well it works once I get it into PS or something?

I'll let ya know tomorrow how things turn out, as I'm re-rendering
all the camera shots I want to show with this new info involved.

And thanx again also............md :)


.

[Shaun_michael]

Hi again,


That all sounds pretty cool. So let me get this straight in my head. You used some non-rendered objects as focal points for the camera? So presumably if you animated these objects you'd be able to achieve a focus oull effect. How does this actually affect the amount of depth of field you have? Does Vue actually adjust this depending on the focal length you set for the camera or is the amount in-focus infront of, plus behind your object on any given frame the same no matter?

I'm not knowledgeable about z-buffers so much, so was curious. One generates the z-buffer and save this out as a sequence (or perhaps embed it in RLA/RPF files), but how does the z-buffer generate the levels of black, white and grey. I know that nearest the camera is black, and furthest is white, but what governs the rate at which it changes from one to the other.

In the lenscare plug you can control this and adjust z-buffer itself, as well as what you can do natively in Combustion to any image in terms of contrast/brightness, gamma, gain etc, and then feed this output to the plugin - this is what makes it so powerful, but I imagine on an image by image basis PS can do this also.

In answer to your question about sequences in PS, if PS can batch process files based on a predefined setting then I guess you can run a sequence through it, but you would need to run two sequences through it - your colour and z-buffer sequences synched.

I'm no PS expert, so I don't know if you can keyframe stuff, but I'm guessing that without a timeline that wouldn't be possible, but someone may know better than that - presumably that's what AE is for.

To make things even more realistic I would certainly bear in mind your camera focal length. Rendering at focal lengths like 25mm and below would tend to give you a greater depth of field in its 35mm stills camera equivalent. The 50mm mark is the 35mm equivalent of what the human eye see in it's field of view (44mm to be exact I think - although it may be 42mm - been a long day), so aiming for the sort of depth of field running say, f5.6, which you can look up on DOF tables, will give you a very natural, human-like feel to the renders. Heading into the telephoto range, as well as flattening the picture will also decrease your depth of field. Of course in stills photography you can compensate somewhat with exposure times and small aperture settings to give you back more depth of field, but this may look slightly unnatural.

Hope that answers some of your questions, but I think I posed more than I answered.

Sorry its taken so long to reply too, had a long day at work, and its freezing over in the UK right now, so I wanted to warm up first!

Cheers,

Shaun

[mdunakin]

OK, first off, now you have me wondering? LOL
Maybe I'm not on track at all, though it sure looks and feels like I am :)
But, since I so far haven't yet tested out any of this stuff in PS yet, I really have no clue?
That all sounds pretty cool. So let me get this straight in my head. You used some non-rendered objects as focal points for the camera?
Yes.
I was basically faking the using of a Null Object in LW effect.
So presumably if you animated these objects you'd be able to achieve a focus oull effect. How does this actually affect the amount of depth of field you have? Does Vue actually adjust this depending on the focal length you set for the camera or is the amount in-focus infront of, plus behind your object on any given frame the same no matter?
Yes, from the way I understand it to work, once you set the camera focus point with
an object, it claims that it doesn't change the camera's actual focal point, but ONLY
the camera's "focus point" and thus, adjust automatically for you, which I checked
to see and each target really did change the focus point of the camera,
so I assume it works if you animated each target.
In otherwords, to me, from what I can see, the DOF "is" adjusted for you automatically
if the target object is moved or animated or if you move/animate the camera it will
still keep the focus point on that target object, which to me sounds
like it's doing the DOF bit as one would expect, right?

I only have a few more renders to go and then I can start playing around with the DOF bit.
As it is, I can do some amount of tricks with the stock PS for getting DOF effects, but that
site you posted does have a demo to download and try out. And if I like what it does and it
does it better then what I can come up with on my own, then I'll just go ahead and order that puppy.
But I have some truely expert friends in these areas of PS and Illustrator type apps,
so I'll be asking them what they think on this subject before I spend any money.

Anyway, hope I answered some of your questions right?
And also, thanx for helping me here :)
And just so you know, I am still learning this and figuring things out as I go :)
And I'll of course, update with anything I create and/or come up with.

Edit: BTW, I forgot to mention, that from the way I see it and the way I read it,
you can still have the camera be targetting an object, (or tracking) say and animated
object in the scene and still be able to have the Focus Point be targeted to it's OWN
target seperate from the camera and also can be animated, thus giving you extreame
control over tracking as well as targetting a Focus Point for the DOF.

Here's a sample showing what the Z Buffer looks like and you can see by the little red
dot I put in there, approximately where my Target Object is for the Camera Focus Point.
My little red dot, is just a guess, as it's hard to tell where I really
have the object placed from this image.

http://www.md-arts.com/Z_BufferFocusPoint.jpg

.............md :)
P.S. I used a simple plane, as that was the smallest amount of polies object I could find.
Since they don't have Nulls.

.

[Shaun_michael]

Well that all sounds rather cool.

Hmmm, still wondering though.

In my mind, from a still, film and video point of view, depth of field is more than just having one object sharp and those infront of, and behind, soft. There is a region, approx one third of the total depth of field in front of the point of focus and two thirds behind the point of focus that is in what's known as "axxeptable focus", and it's this total distance that defines the depth of field. Again, in the real world, this distance can be increased and decreased depending on aperture settings/exposure time or focal length, and I'm curious (because I simply dont know), whether Vue, or LW for that matter, can alter it's z-depth output to to replicate this...

... perhaps it's just a matter of adjusting the z-depth in PS/Combustion etc to change the levels, and thus how fast white turns to black, and clamping certain level ranges to simulate that region of depth of field.

Wish I knew more about z-depth.

Looking forward to seeing and hearing about your findings. Am enjoying your tressel thread, and many others as it's an invaluable learning source, and an excellent point of contact for bouncing ideas.

Cheers,

Shaun

PS - there look like some setting in Lenscare that might do this within the plug (thats what it appears to be), but haven't experiemnted fully with them yet. Its a good plug, but like you say, you may have some expert advise that points to the same efect from within PS etc.

[mdunakin]

OK, here' my first attempt with the Z Buffer and PS and my image.
First off, I should of known better then to render out a slew of images without
first testing this stuff, since I found out that ONLY the color button you select
to save, gets saved from Vue, and that all that other info doesn't get
saved into the PSD unless you save them seperately.
In otherwords, you need to save out the Color and the
Alpha and the Z Channels all as seperate images.

I downloaded the demo of that Lenscare plugin, but still haven't tried it.
I wanted to see what PS would do on it's own first.
First off, that old PS trick that you can do, doesn't really do anywhere near the justice
that a rendered out Z Buffer image can do, which is to just make a gradient and use
that as an Alpha Channel and then use that to make the DOF effect with PS's Lens Blur.
Using the rendered out Z Buffer image makes all the difference in the world.
You just need to remember to save the thing out.
Though I haven't checked into any info on the G Buffer deal in Vue.
Will find out what that is later, no time today.

Um, anyway, what I did, was to save out both the Color and the Z channels and then load those into PS.
I then copied and pasted the Z image over to the Alpha Channel for
the color image or you can create a new Alpha channel if you needed to.

Then using the Lens Blur filter I just selected the Alpha channel I just pasted down, for my bluring.

Lens Blur allowed me to adjust where the center point could be set to as well
as how much depth and bluring I wanted, just like a real plugin, or something LOL

But, the point here, is that by using Vue's Z rendered image, you can see that the DOF works more real, i.e. the trestle tops are closer to the camera and thus more blury
then the ground on the under side of the trestle that's also down at a lower level,
thus making the blur be different like it really would be,
instead of just bluring out everything the same.
And the dirt/rocks slowly and smoothly blend
the blur as they get more into the distance.
I was trying to draw the eye to the top left corner of the trestle and tracks where
the grass is and the slope of dirt goes down into the wash, and that was my
fucus point set up in Vue and it worked out pretty good in post too.

Really makes all the difference in the world using that Z image!
http://www.md-arts.com/_desert_trestle_wash_eco_03_TrkTDwDOF_09.jpg

EDIT: And the Z Buffer image I sued to create the DOF in PS.
http://www.md-arts.com/_desert_trestle_wash_eco_03_TrkTD_09z.jpg

And I can of course adjust things more if I wanted to, but I kinda like the
amount of blur (DOF) I have going on right now, of not too much and not tooo little.
IMO, of course :)

Give me your opinions on this, as I want to hear some
opinions before I do any other renderings all over again LOL

....................md :)


.

[Shaun_michael]

Hi,

That looks neat. I love the lighting and texturing and the landscape is looking really realistic.

OK, on the DOF. As I mentioned in an earlier post, in my mind depth of field is a range of an image on the z-axis where the image appears in focus. This will be made up of you point of focus which will be pin sharp, and an area infront of, and behind, which will be in acceptable focus i.e. your eyes will not be able to discern that it is not pin sharp because our eyes simply aren't that good. However, this will always depend on the size of the print, and the distance from which you view it (oh, and those pesky circles of confusion).

Having said all that, this is only relevant to real life photography where film speed (grain size), and CoC do change size. In CG of course, the pixel is always the same size. But to create a realistic DOF effect, we would still expect to see a "band" of the image in-focus and infront and behind that, the image would travel out of focus at the speed determined by the aperture setting.

If you look at your z-buffer, you'll notice that it travels from white to black. To create a band that's in focus, we need it to start off black/grey, travel to white at and around your focal point, then back to grey then black as you travel past your area of focus into the distance.

Lenscare doesn't give you this exact control, and doesn't need to as you apply it in a compositing application that will allow you to take the z-buffer output from lenscare, and apply curve changes to it.

I've done a quick render of one of the sample scenes in Vue 5I, taken the z-buffer also and applied them in Combustion with lenscare. I then adjusted the curve of the B&W z-buffer to create a non-linear z-buffer curve to give me that band of white around my point of interest, and this area in-focus in the depth that I am looking for. I can adjust the curves to give me virtually any area of the image in focus without having to add objects to my render, and of course, as my curves are animatable, I have full control over the way the focus changes. The third of the three images is an output from lenscare that shows me exactly what portion of the image (the whote portion) is sharp, allowing me to have instant feedback when adjusting the portion in focus, and also the depth (amount of z-axis included) within my DOF effect

Just in case you weren't aware. If you render out and .rla or .rpf file, this will automatically contain, amongst other picture info, your z-buffer, which means you dont have to render a seperate z-buffer sequence if you dont want to. This assumes your comping software reads rpf/rla files.

Your image looks great. The DOF is pretty minimal (as can be seen from your z-buffer grey values), so you're not going to get a tremendous effect unless you introduce more contrast to the z-buffer image. With composition it's pretty general to draw the eye to the strongest element in your scene with you DOF. In your image, along with its very strong perspective, your rails draw the eye much stronger than the landscape does - especially as your landscape is low contrast, as you'd expect in the environment you've created, so the DOF doesn't jump out. Its nice and subtle - but doesn't highlight anything particularly.

If you were thinking of using DOF as a tool to draw the eye, and a compositional aid, as well as adding real-world effects, it will now depend on the overall composition of the shot. Try to avoid areas that will draw the eye away from the area you want people to see. If it's the landscape you want to showcase in a shot, then you'd need to weaken the rails by changing your POV. If its the overall scene you want to highlight, you would of course use the perspective effect from the rails to lead your eye to the landscape, and if it were the rails, you would make them prominent in the F/G and use a very small area of DOF that highlighted them, and dropped the landscape out of focus.

I hope all that makes sense in some way. Of course, as with all art, my view will not be the same as someone elses, and only you know exactly what you're trying to achieve. If you can objectively ask yourself "Does this image show me what I want it to show me, or am I being drawn elsewhere in the image" and look at the answer, then you'll not go wrong.

Sorry, I sound like I'm lecturing. I must stop!

Looking forward to seeing more of your work.

Cheers,

Shaun

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.