limited DOF

[Crazy Homeless Guy]

i am going to try and open a small discussion about Depth of Field (DOF). i am not a photographer, and my knowledge of how DOF is limited, but expanding.

 

it seems like more and more people are using DOF in their images. it can have a powerful effect on images when used correctly, but i think it is difficult to use correctly, myself included. people that use it are often criticized for their use of it, with no real explanation of why their use of it is poor other than it controls the image.

 

the term 'photoreal' is thrown around a lot, so photography makes an ideal place to start dissecting DOF.

 

i found this image in a magazine several weeks ago, and it made me start thinking about how a similar effect could be used successfully in rendering, without your eye getting stuck on it. i look at this image, and i am very aware of the DOF, but it is not what i am concentrating on.

 

http://www.cgarchitect.com/vb/attachments/16181d1157031160-hotel-swimming-pool-car.jpg

 

i have an old Canon AE1 that i used to mess around with, but haven't touched for a couple of years. i don't yet have a digital SLR, although one is sounding better everyday. this is where Vray1.5 and its new physical camera comes in handy. i still have the power the computer provides of making things perfect, but can experiment using a camera that represents the way a real camera works.

 

i set this scene up for experiment purposes. a series of boxes in the background that could be representative of buildings across a busy street, the foreground box might be representative of a new library on a site the size of a city block.

 

for the overall shot i used a Vray physical camera with settings that are commonly used for as architectural visualization, i forgot to write them down, but assume it is about a 35mm lens maybe a 6 f-stop, 30 shutter speed on 200 ISO.

 

http://www.cgarchitect.com/vb/attachments/16920d1161461341-limited-dof-35.jpg

 

this shot is purely a reference for the time being. i am currently experimenting with a higher mm lens, because that is the type of lens used commonly for this type of effect in the real world. again, i have very little knowledge in lenses and photography, so if i am wrong on something, please speak up and tell me. no misinformation.

 

 

the second shot was created by switching the camera to a 120mm lens and starting to play with the f-stop, and shutter speed. i am at home now, and was working with this at the office, so i am writing my settings down from memory.

 

http://www.cgarchitect.com/vb/attachments/16921d1161461341-limited-dof-dof-16-20.jpg

 

120mm lens

16 f-stop

20 shutter speed.

400 ISO

as expected, the field of view (FOV) is a great deal tighter as expected, and the image has a good DOF using these settings. i can't really say much more about it than that. pretend the box on the left is the side of the library, and you are looking towards the other side of the street. ...though you are not looking at the other side of the street. i need to make that clear do justify the settings on the third shot. if you were trying to look at the other side of the street, the the extremely limited DOF i am about to do would be a lost cause.

 

the third shot was created by making the following adjustment to the camera settings....

 

http://www.cgarchitect.com/vb/attachments/16922d1161461341-limited-dof-dof-1.4-3000.jpg

 

120mm lens

1.4 f-stop

3000 shutter speed.

400 ISO

as you can see, all that i changed was the f-stop, and the shutter speed. these changes caused a DOF change that greatly impacted how you look at the image. whether it is better or worse is up to the viewers decision, but it definitely brings up some interesting thoughts on what you could do by playing with these settings on renderings.

 

the downside of all of this is time cost. the initial overall view only took 20 seconds to render. much like a traditional Max depth blur, the render engine is put under extreme duress when trying to calculate and produce an image with blurring. my 20 second render time went to 5:40 for the second image, and all the way up to 8:20 for the third. because of this i am not sure how useful these experiments will be in my overall work flow. they would only be valid on images that you are willing to start, walk away, and not come back until the next day. even then, you are not guaranteed the image will be done, and this is true regardless of the personal PC you are using.

 

it just further goes to show how much more powerful a camera and a set of lenses are compared to th e computer. it is not really a fare comparison because the computer has to worry about simulating all of the aspects of a scene, and a traditional camera gets to focus solely on capturing that scene. the technology used lenses has a pretty fixed set of rules. they have evolved over the years, but still the same basic concept of how they work.

 

point and shoot digitals try to mimic a wide variety of lenses, but they are completely inferior to the power of real lenses. the next generational step in digital photography imo might involve the camera understanding 3d space, rather than interpreting 3d space as a 2d image. i am sure a technology like this already exists, or is being developed, but it is far from working its way into our hands by form of a point and shoot camera from NewEgg.

 

blurring in renderings is quite frustrating because of the time requirements to execute it. the only fast solution we are given is to render a Z-depth pass, and then apply the blurring manually in PhotoShop, or a compositing app of our choice. and i suspect most this is how most of the DOF blurs we see posted. this is how i have done it in the past. the problem is that we control the set of rules, and not the lenses. we apply blurring based on how we feel it should be, and not by what the camera controls will allow us. i guess this has its upsides, and downsides. the biggest downsides are that the blur is not as elegant as it is when created using a camera effect.

 

i know there is a depth blur plug-in for PhotoShop that i have not played with. maybe i should mess around with that, and see what it is capable of. i am guessing it can simulate some of the lens effects than PhotoShop can by what is included.

 

DOF is something that we deal with everyday with the human eye. the human eye is very sophisticated and focuses so quickly when looking around that we don't even notice it. we take it completely foregranted. it is the way it has been, and it is the way it will be. when their are no changes to something it is easily to take that thing foregranted. when using DOF in photographs or renderings, we are trying to freeze a moment of time that we are not used to noticing. ...or freeze it in a way that typically don't think about.

 

one of the interesting side effects of limiting the DOF is how it effects the viewers sense of scale about the objects int he image.n it often makes the object in the images look toy like. again, whether this is good or bad is completely up to the viewers tastes.

 

i think the reason objects become toy like is related to how the human eye works in our everyday experience. when we are just looking around we really don't notice any DOF blurring. it is not until we hold an object close to our eye, and focus on it do we start to notice how blurred everything is behind the object you are holding an inch or two from your eye. i think we might relate that sense of scale to the images we see of large objects that have a similar sense of DOF blurring. are minds are simply reacting to the scale we are used to seeing when things look like this in real life. sure, DOF is present at all times when we are looking around, but i am arguing that it is only noticeable at a few times.

i think experimenting with fish lens settings will be my next experiment with Vray's physical camera. i also want to study how wide a DOF is when using a standard 28 or 50mm lens. this will give a true answer to how we are using DOF wrong in renderings

i am reading Ansel Adams set of book's right now, The Camera, The Negative, and The Print right now. i am sure these will provide more valuable information over the next couple of months. i am also going to spend some time browsing around the web for resources.

 

if you made it all the way to the end of my post without leaving, thanks.

 

i am posting this in the general forum, because it is not necesarrily about the Vray physical camera, but it does turn into a discussion about that, i will have it moved to the Vray forum to keep the board neat and tidy.

 

.

 

NP : KEXP Quilty 3000

[martin walker]

GOOD STUFF.. !

 

 

you also win the "longest post ever" award

[Koper]

Yeah, good stuff! To those who see the length of the 1st post, yeah it’s long but I think very interesting and good reading!

 

Hey Crazy HG, I partially agree with you that dof gives a toy like feeling (if it is small). As small children, when we played with toys we really focused in on those particular toys we played with, and that is exactly what dof is, giving focus to the subject mater.

 

We recognize dof much easier with macro photography because the area in focused is much smaller and sharper, its easy to identify. This link shows this.

http://www.dpchallenge.com/image.php?IMAGE_ID=59606

 

But i think your looking at dof the wrong way. Rather think of dof as to how much is in focus, and not how much blur there is. Landscape photography uses a very big dof most of the time so that everything is in focus. This link demonstrates this.

http://www.dpchallenge.com/image.php?IMAGE_ID=81455

[IC]

I'd rather you didn't move this post as I think it's general enough (with other render engines having physical camera attributes) to remain here.

DOF is a difficult thing to get right so it would be great to have more technically adept people like yourself posting input.

 

I have an old SLR and I also spent some time experimenting with the focus settings to get a feel for mimicking them but it's not quite that easy. Sometimes (in CG) it works and sometimes it doesn't.

[Sawyer]

I think it is worth mentioning that the first image does not use DOF. It has a focal blur.

 

Typically for exteriors you have a wide view and a far away target. This limits the amount of blur you will have because once you get a certain distance away from hte camera objects usally are in the same focal range. Something at 20' away is usually in the same focal range as something 100' away. I think its typically around 16' away depending on your lens but once you hit that point the camera sets for infinity. An extremly close object will blur.

[martin walker]

DOF as an artistic effect is great, but its use in a "photo realistic" scenario is very limited. Its uses makes arch vis look like tiny models, and thats exactly how it is in nature. For example....stand on a hill and look over a city scape....its all perfectly in focus, now close one eye and move your other eye to within 5 cm of your mouse and either it will be blurred, or your mouse mat will !

 

A big part of it is human binocular vision, but the main part is the fact that the human eye is designed to have a huge focal range, but not great at macro tasks

 

I really like DOF, but in our line of architectural visualisation I think its a slightly twee effect that is very limited, and when used incorrectly it can destroy and otherwise compatent visual

[Koper]

here is something interesting to read to those who want to create film like quality renderings, stills and/or animations.

 

everybody knows pixar and this is how they do it, see it.

 

some stuff are a bit of a higher grade but it is still good reading

 

Enjoy

 

http://graphics.pixar.com/DepthOfField/paper.pdf