Digital Art Zone

 
   
4 of 4
4
Pros and Cons of Carrara DOF Vs. Postworked DOF
Posted: 11 February 2013 08:00 PM   [ Ignore ]   [ # 46 ]
Active Member
Avatar
RankRank
Total Posts:  647
Joined  2003-10-09
thoromyr - 11 February 2013 07:36 PM

Except where its not smile

He suggests that the Carrara built in DOF is okay for animation when I disagree. For animation you would want to use post working, preferably something like after effects, though even photoshop batch mode would work. ...

Steve K - 11 February 2013 07:22 PM

In any case, if there is no camera motion or animation, you can render only the two frames and use a crossfade in the video editor (as I mentioned a couple of messages back).  Admittedly this is fairly limiting ...

Your specific case could, in principle, be handled by one image plus a depth map and some work in post. Not something I’ve done, but that is exactly the kind of thing post work is good for (doing blur for different depths in the map in different frames).

OK, thanks for the elaboration.  I do have a fairly recent version of AE and so I am interested in your comments.  OTOH, my perspective is seriously warped by my participation in the 48Hour Film contest, where “time is of the essence”, as the lawyers say.  So a couple of renders in Carrara and a crossfade in the video editor is OK.  Getting a third app involved is not generally a plus.

But yes, most animations are not done in two days.  Probably almost none.  So I appreciate the tips.

Profile
 
 
Posted: 11 February 2013 08:28 PM   [ Ignore ]   [ # 47 ]
Active Member
Avatar
RankRank
Total Posts:  381
Joined  2008-08-29
Steve K - 11 February 2013 08:00 PM
thoromyr - 11 February 2013 07:36 PM

Except where its not smile

He suggests that the Carrara built in DOF is okay for animation when I disagree. For animation you would want to use post working, preferably something like after effects, though even photoshop batch mode would work. ...

Steve K - 11 February 2013 07:22 PM

In any case, if there is no camera motion or animation, you can render only the two frames and use a crossfade in the video editor (as I mentioned a couple of messages back).  Admittedly this is fairly limiting ...

Your specific case could, in principle, be handled by one image plus a depth map and some work in post. Not something I’ve done, but that is exactly the kind of thing post work is good for (doing blur for different depths in the map in different frames).

OK, thanks for the elaboration.  I do have a fairly recent version of AE and so I am interested in your comments.  OTOH, my perspective is seriously warped by my participation in the 48Hour Film contest, where “time is of the essence”, as the lawyers say.  So a couple of renders in Carrara and a crossfade in the video editor is OK.  Getting a third app involved is not generally a plus.

But yes, most animations are not done in two days.  Probably almost none.  So I appreciate the tips.

Well, as I said I haven’t done that—but I understand the principle. I don’t have AE at all, but I have used photoshop to batch process frames. It wasn’t until a previous DOF discussion on the forum that I seriously tried Carrara’s DOF capability. Before that I’d tried the quick Carrara DOF and had been very disappointed, and then turned off by the slow render time of the ray traced variation.

Even though I didn’t set out to do so, the scene I created as a result of the discussion was very good at showing the weakness of relying on depth maps. All I wanted was something that naturally had DOF, not from being forced. Hence using a really great range of depth in the image.

As it turns out, the greater the range in depth in your image the more useless depth maps (Carrara’s “fast” option, or doing a depth map pass and using post processing) are. With a range of only 0 to 255 there just isn’t much granularity. If your image has one spot that “goes all the back” (e.g., to the sky) then your foreground will all be a single depth as the 255 gets stretched to whatever value carrara uses for “indefinitely far away” with some middle ground getting the grey in-between.

To alleviate that, even if the final render will include the background, mask out the background with an object at whatever depth you want as the limit for a depth pass. It requires a separate render with the object, but you can then plug that into whatever post processing workflow you have. From what I understand of AE, you should then be able to select the depth of focus at arbitrary frames and use the depth maps with that. Doing the same with Carrara’s “quick” version would require setting the focus depth at those frames, then a full render with a single threaded DOF effect and then reviewing the result to ensure it was good. In AE, the DOF effect is multi-threaded and if you need to tweak the focus depth for a frame its just a fast recalculation rather than a re-render.

However, if I was doing animation and wanted DOF I would *only* do it through post processing with a depth map. Even if you were fine with taking the first result of the Carrara rendered DOF, it is just too slow (even the “quick” version thanks to being single threaded, the time adds up with the number of frames). I would just ensure that I had usable depth maps smile

 Signature 

Swamp Living Never Felt So Good

Profile
 
 
Posted: 11 February 2013 09:44 PM   [ Ignore ]   [ # 48 ]
Addict
Avatar
RankRankRankRank
Total Posts:  4317
Joined  2006-08-27
thoromyr - 11 February 2013 06:42 PM
evilproducer - 11 February 2013 01:45 PM

The built in DOF may work better for animations where the viewer’s eye is more focused on the animation and may only be peripherally aware of the effect.

I really don’t see how. If you are using the ray traced render it will be much slower than using after effects and for animation the quality is less important than with a still—its the overall effect that matters. With a still, that’s all you’ve got. The viewer sees that same frame, and keeps seeing it, and seeing it, and seeing it without change instead of only seeing it for a fraction of a second.

And if you don’t use the ray traced DOF then it’s quick, but what you get is a pre-dialed effect so you lose any control like you might use in post work.


Not sure where you picked up on the ray traced thing. I never said ray traced DOF would be good for an animation. I said that the built in DOF may work better in some animations. Since the default setting for DOF in Carrara appears to be the post effect version, that’s what I’m taking about.


Where you may have issues using postwork DOF for an animation is if you have something in your scene such as splats or billboards. I haven’t used all the render passes available in Carrara, but the ones that I am familiar with don’t seem to respect alphas. I can see this causing problems in cases where there are billboards to speed render times.

 Signature 

I find it somewhat liberating not to be encumbered by accuracy.

Profile
 
 
Posted: 11 February 2013 11:48 PM   [ Ignore ]   [ # 49 ]
Member
Avatar
Rank
Total Posts:  106
Joined  0

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Profile
 
 
Posted: 12 February 2013 01:46 AM   [ Ignore ]   [ # 50 ]
Power Member
Avatar
RankRankRank
Total Posts:  1059
Joined  2007-10-15
Bagboy - 11 February 2013 11:48 PM

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Excellent question…

In fact, in many ways it’s a pretty complicated answer.

Here’s the short answer: You tell the “blur-er” application the name of the depth pass image and it will use that to do the blurring

And here’s the long answer: In its simplest form, Depth of Field is just an indication of how “in focus” each part of an image is. So if you have an image and want to make it look like there are parts of the image that are out of focus due to the DOF effect, somehow you need to tell your “Blur-er” application how far each element of the image is from the rendering camera.

In fact, you need to tell the “Blur-er” how far each part of each underlying object in each pixel of your image is from the lens of the rendering camera. You do that by generating an image that, in this case, is basically a greyscale image where the brightness level of each pixel of that greyscale image reflects the relative distance of that part of the underlying object from the camera lens. That is the depth map image you generate in the render pass.

So in effect the greyscale image becomes a code where the brightness level of each pixel gives the relative distance between the underlying object and the rendering camera.

Now in the application you’re using to determine how to blur your image, it needs to know how much to blur each pixel. There are many ways to tell it that information, but one way is that the Blur-er application will allow you to specify the greyscale “depth” image that you got as one of the render passes when it asks you how it should do the blur. So there will be some way to say to the “Blur-er” application “okay, I want you to blur this image based on this greyscale I generated as one of the render passes”.

Now, that doesn’t mean that you NEED a render pass to generate a depth map. All you need is a a greyscale image where the brightness of each pixel reflects a relative distance. In fact, in many cases you can very easily generate your own depth maps. This will give you a lot more artistic control, and often give you better and faster results.

How? Well the simplest depth map is simply a black-to-white gradient drawn from the bottom of the image to the top of the image. Bottom of the image is the floor under your feet, and top of the image is stuff off in the distance. Or, if you have a character standing in front of the camera you can cut out that the character part of the image and paint it full black (which means it’s close to the camera), and have the rest of the image be blurred based on your simple gradient.

Or you can use your artistic creativity to generate a freestyle depth map, using spherical gradients or whatever you want. The sky is the limit.

However, in real life the DOF effect is really quite complicated, and is nothing like a simple blurring. DOF is due to the physical characteristics of cameras and lenses, and can be quite complicated if you want to reproduce those characteristics. And often it is very important to reproduce them accurately, because humans are very used to seeing DOF and related effects, so they are very attuned to what they expect.

In practice, the blurriness of the image due to the DOF effects is a function of how “zoomy” the camera view is (how much zoom/magnification the lens has and how close it is to the subject) as well as how dark the scene is, since it is a function of how open the aperture is. Zoomy-er lenses that are close to the subject in low light levels generally have the most blurry (shallow DOF) images.

But the actual visible effect is also a function of other things, and it’s not just a simple blur. It depends on the shape of the aperture, for example, which can affect the overall look of the blur and its highlights. There are a number of related lens effects that can be considered connected with, and related to, DOF. Which is one reason why I suggest people don’t rely upon simple DOF blurs that are designed to merely give a simple approximation to something that is often very complicated and interesting. Very often it’s those subtle little complications that make viewers subconsciously suddenly realize “hey, that’s REAL !!” and they instantly believe your premise.

If you have one, that is…   

 

Profile
 
 
Posted: 12 February 2013 03:45 AM   [ Ignore ]   [ # 51 ]
Power Member
Avatar
RankRankRank
Total Posts:  1059
Joined  2007-10-15
thoromyr - 11 February 2013 08:28 PM

As it turns out, the greater the range in depth in your image the more useless depth maps (Carrara’s “fast” option, or doing a depth map pass and using post processing) are. With a range of only 0 to 255 there just isn’t much granularity. If your image has one spot that “goes all the back” (e.g., to the sky) then your foreground will all be a single depth as the 255 gets stretched to whatever value carrara uses for “indefinitely far away” with some middle ground getting the grey in-between.

You sure about all that?

Profile
 
 
Posted: 12 February 2013 07:29 AM   [ Ignore ]   [ # 52 ]
Active Member
Avatar
RankRank
Total Posts:  381
Joined  2008-08-29
evilproducer - 11 February 2013 09:44 PM

Not sure where you picked up on the ray traced thing. I never said ray traced DOF would be good for an animation. I said that the built in DOF may work better in some animations. Since the default setting for DOF in Carrara appears to be the post effect version, that’s what I’m taking about.

Just part of giving a full answer. And I addressed the lack of advantage of the ‘quick’ version.

evilproducer - 11 February 2013 09:44 PM

NWhere you may have issues using postwork DOF for an animation is if you have something in your scene such as splats or billboards. I haven’t used all the render passes available in Carrara, but the ones that I am familiar with don’t seem to respect alphas. I can see this causing problems in cases where there are billboards to speed render times.

There are no billboards, etc., in the example image I posted. I’ve already explained with depth maps don’t work for it and how that could be worked around (as long as you are only after blur in the foreground). Notice that all the images posted have a near background.

DOF in post is a good idea when:
  * speed matters
  * control matters

built in ray traced DOF is a good idea when:
  * accuracy matters
  * large depth range is involved

For the latter, the interest is primarily in effect. I believe it was one of yours earlier where you described editing the depth map. More power to you, post work allows you to have more control over it. And, in practical terms, a large depth of range does not appear to normally be involved. Its no wonder people prefer doing it in post.

 Signature 

Swamp Living Never Felt So Good

Profile
 
 
Posted: 12 February 2013 07:30 AM   [ Ignore ]   [ # 53 ]
Active Member
Avatar
RankRank
Total Posts:  381
Joined  2008-08-29
Bagboy - 11 February 2013 11:48 PM

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Photoshop has a lens blur function for which you load the depth map. Outside of that it can be used as a mask for blurring. That’s the short of it anyway. If you google for it you should be able to find tutorials (which is what I did when I started on exploring this from the previous thread, but I don’t recall any URLs).

 Signature 

Swamp Living Never Felt So Good

Profile
 
 
Posted: 12 February 2013 04:39 PM   [ Ignore ]   [ # 54 ]
Power Member
Avatar
RankRankRank
Total Posts:  1059
Joined  2007-10-15
thoromyr - 12 February 2013 07:29 AM

DOF in post is a good idea when:
  * speed matters
  * control matters

built in ray traced DOF is a good idea when:
  * accuracy matters
  * large depth range is involved

For the latter, the interest is primarily in effect. I believe it was one of yours earlier where you described editing the depth map. More power to you, post work allows you to have more control over it. And, in practical terms, a large depth of range does not appear to normally be involved. Its no wonder people prefer doing it in post.

You sure about all that?

Profile
 
 
Posted: 12 February 2013 05:20 PM   [ Ignore ]   [ # 55 ]
Member
Avatar
Rank
Total Posts:  106
Joined  0

Hey JM!

Thanks for the in depth explanation!  There’s a lot to chew on here.

I really appreciate this!

JoeMamma2000 - 12 February 2013 01:46 AM
Bagboy - 11 February 2013 11:48 PM

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Excellent question…

In fact, in many ways it’s a pretty complicated answer.

Here’s the short answer: You tell the “blur-er” application the name of the depth pass image and it will use that to do the blurring

And here’s the long answer: In its simplest form, Depth of Field is just an indication of how “in focus” each part of an image is. So if you have an image and want to make it look like there are parts of the image that are out of focus due to the DOF effect, somehow you need to tell your “Blur-er” application how far each element of the image is from the rendering camera.

In fact, you need to tell the “Blur-er” how far each part of each underlying object in each pixel of your image is from the lens of the rendering camera. You do that by generating an image that, in this case, is basically a greyscale image where the brightness level of each pixel of that greyscale image reflects the relative distance of that part of the underlying object from the camera lens. That is the depth map image you generate in the render pass.

So in effect the greyscale image becomes a code where the brightness level of each pixel gives the relative distance between the underlying object and the rendering camera.

Now in the application you’re using to determine how to blur your image, it needs to know how much to blur each pixel. There are many ways to tell it that information, but one way is that the Blur-er application will allow you to specify the greyscale “depth” image that you got as one of the render passes when it asks you how it should do the blur. So there will be some way to say to the “Blur-er” application “okay, I want you to blur this image based on this greyscale I generated as one of the render passes”.

Now, that doesn’t mean that you NEED a render pass to generate a depth map. All you need is a a greyscale image where the brightness of each pixel reflects a relative distance. In fact, in many cases you can very easily generate your own depth maps. This will give you a lot more artistic control, and often give you better and faster results.

How? Well the simplest depth map is simply a black-to-white gradient drawn from the bottom of the image to the top of the image. Bottom of the image is the floor under your feet, and top of the image is stuff off in the distance. Or, if you have a character standing in front of the camera you can cut out that the character part of the image and paint it full black (which means it’s close to the camera), and have the rest of the image be blurred based on your simple gradient.

Or you can use your artistic creativity to generate a freestyle depth map, using spherical gradients or whatever you want. The sky is the limit.

However, in real life the DOF effect is really quite complicated, and is nothing like a simple blurring. DOF is due to the physical characteristics of cameras and lenses, and can be quite complicated if you want to reproduce those characteristics. And often it is very important to reproduce them accurately, because humans are very used to seeing DOF and related effects, so they are very attuned to what they expect.

In practice, the blurriness of the image due to the DOF effects is a function of how “zoomy” the camera view is (how much zoom/magnification the lens has and how close it is to the subject) as well as how dark the scene is, since it is a function of how open the aperture is. Zoomy-er lenses that are close to the subject in low light levels generally have the most blurry (shallow DOF) images.

But the actual visible effect is also a function of other things, and it’s not just a simple blur. It depends on the shape of the aperture, for example, which can affect the overall look of the blur and its highlights. There are a number of related lens effects that can be considered connected with, and related to, DOF. Which is one reason why I suggest people don’t rely upon simple DOF blurs that are designed to merely give a simple approximation to something that is often very complicated and interesting. Very often it’s those subtle little complications that make viewers subconsciously suddenly realize “hey, that’s REAL !!” and they instantly believe your premise.

If you have one, that is…

Profile
 
 
Posted: 12 February 2013 05:23 PM   [ Ignore ]   [ # 56 ]
Member
Avatar
Rank
Total Posts:  106
Joined  0
thoromyr - 12 February 2013 07:30 AM
Bagboy - 11 February 2013 11:48 PM

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Photoshop has a lens blur function for which you load the depth map. Outside of that it can be used as a mask for blurring. That’s the short of it anyway. If you google for it you should be able to find tutorials (which is what I did when I started on exploring this from the previous thread, but I don’t recall any URLs).

Thanks Thoromyr!

I’ll have a look out for that lens blur function.

Profile
 
 
Posted: 27 February 2013 11:36 PM   [ Ignore ]   [ # 57 ]
Power Member
Avatar
RankRankRank
Total Posts:  1792
Joined  2008-01-28

great thread Evil, thanks for starting it

 Signature 

http://andrewfinnie.blogspot.com.au/

Profile
 
 
Posted: 28 February 2013 04:21 PM   [ Ignore ]   [ # 58 ]
Addict
Avatar
RankRankRankRank
Total Posts:  4317
Joined  2006-08-27
Bagboy - 12 February 2013 05:23 PM
thoromyr - 12 February 2013 07:30 AM
Bagboy - 11 February 2013 11:48 PM

I’ve never understood what one did with the DOF pass image to create, let alone control, the blur in an image editor. 

Any hints for a clueless one?

Photoshop has a lens blur function for which you load the depth map. Outside of that it can be used as a mask for blurring. That’s the short of it anyway. If you google for it you should be able to find tutorials (which is what I did when I started on exploring this from the previous thread, but I don’t recall any URLs).

Thanks Thoromyr!

I’ll have a look out for that lens blur function.


It’s under blur and it seems to work quite nicely.

 Signature 

I find it somewhat liberating not to be encumbered by accuracy.

Profile
 
 
Posted: 04 August 2013 05:31 PM   [ Ignore ]   [ # 59 ]
Addict
Avatar
RankRankRankRank
Total Posts:  5324
Joined  2007-12-04

Just an FYI:
I have used Photoshop 7 and then my nephew moved into the other side of the duplex in which I live, and he let me use his Photoshop CS for quite some time. I ended up spending quite a bit of time over there with my handy-dandy usb hard drive optimizing images and such - which helps to explain why I often did all of my effects within Carrara, attempting to go for “No Post Work” images. When we promote our products, here at DAZ, we are supposed to use images that haven’t been altered by post filters and such - so that the customer knows what to expect, out of the box.

Photoshop CS is still a bit out of my reach. Sure, I could go without allowing my family to eat for a month and buy it anyways… but I opt to not torture my loved ones in such a way. So I’ve been using TheGimp, which truly is an outstanding piece of software, and it’s free! I’ll never go without it in my toolbox. But I recall Rogue Pilot mentioning techniques he’s used in PaintShop Pro several times here and there. So recently, I decided to look into it. PaintShop Pro, many years ago, was the first software I’ve owned that allowed me to use layers for editing. I still have that somewhere around here, and was planning to hunt it down and install it again, but Corel (my version was before Corel took it over) was offering it for their anniversary special at the low, low price of $29.95 USD - and I could get it in it’s X5 (latest at the time) version PLUS the VideoStudo Pro X5 (not the latest, as X6 has just come out at the time) for a low-cost bundle, so I bought that. Already having Sony Movie HD, I looked pretty deep into VideoStudio Pro before adding it to the buy, and sure enough, it has some great feature to bring to the table that I don’t otherwise already have available.

Anyways, I’m still getting used to PaintShop Pro X5. So I’m catching some critique on some of my images that (I know) actually came from my post work attempts, rather than entirely due to my rendering qualities. I know, big deal, right? That’s how I look at it. We’re all friends here - and I don’t mind if I post something that you guys feel is substandard. We all learn together, right?

Well what drew me into PSPro over getting Photoshop Elements (their cheaper version, which is also available in a handy bundle with Premier Elements for a sweet deal), is that it can merge HDR Images from Exposure shots, as explained in Phil Wilkes Advanced Training, and Elements, I was told by a rep at Adobe, does not. I would need to get Photoshop CS for that.

Well I’m finding that Corel has really implemented some great learning aids right into the software - that I find to be superb. Utilizing “Help” in some software doesn’t always get you very far on the ol’ learning curve. Well Corel offers (even via pop-up until you choose to not see the popup again) for $4.99 each, different levels of training that you can add to the interface - which I’m not planning to get. That stuff seems aimed more towards users who really need help getting started in photo editing software for the firsat time - which I think is super-awesome. $5 is really cheap for a really good “Getting Started” video learning aid, and there are several levels from which to choose. So after ‘opting out’, I opened X5 again and just checked the Help menu, which already has an extensive set of “Getting Started” material, which give a handy bit of info on how to access the features you’re looking to use, which then offers a “More Info” button that launches your browser on the subject. Pretty neat!
But Help also includes a guide to your Keyboard Shortcuts, user manual and other learning tid bits.

I haven’t even begun to try VideoStudio Pro yet, but it is advertised to have the same sort of helpful interface as well - something that originally drew me into purchasing Sony Movie HD as well. I like it when I can learn software within that software. That’s what really gave me the idea to include tips and such in my EnvironKits packs.

And I’m sure you all already know about how much I think that Project Dogwaffle Pro: Howler is just Cookies and Cream for Carrara animation endeavors - and stills alike!

Interested in the great line of Corel products for post work? Here’s a link to their friendly, easy to navigate site. Helpful to the end. I like that in a software company. And here’s a video about the Photo/Video suite I bought

Image Attachments
m5580090_photo_video_suite_X5_205x211.png
 Signature 

Dartanbeck @ Daz3d          Check out the Carrara Cafe          ►►►  Carrara Information Manual   ◄◄◄

Profile
 
 
Posted: 04 August 2013 06:08 PM   [ Ignore ]   [ # 60 ]
Addict
Avatar
RankRankRankRank
Total Posts:  5324
Joined  2007-12-04

Oh… and additionally, PaintShop Pro has a handy popup when opening the software where you can get answers to questions you may have, and also download and install many free ‘content packs’ to further your post work experience. Each pack that you may download and install as such, also has a information button where you can access instructions for use of that pack. Now that’s service!

I am also ultimately impressed at how intuitive and easy to navigate their site is. They have a multitude of products that will aid us in our post-working endeavors. Helpful tutorials, written and video, can be fully accessed prior to purchase. I wasn’t really expecting PaintShop Pro to be quite as robust a software that it is. And, sure, you can likely do much more with Photoshop CS. But it’s just amazing what a great value this software truly appears to me to be.

Just hoping to help some who may be looking for options - as I find this to be an excellent choice! wink

 Signature 

Dartanbeck @ Daz3d          Check out the Carrara Cafe          ►►►  Carrara Information Manual   ◄◄◄

Profile
 
 
   
4 of 4
4