Fast lenses, depth of field, and Photoshop CS6

As I understand it, the new CS6 will have a very intriguing tool. It will be able to create depth of field, and do it simply.

This begs the question: who needs an expensive, fast lens?

One might say that "it's not the same as what a lens can do." Perhaps? Perhaps, not.

For example, rather than buy the 70-200 2.8 II IS, one might be better off with the 70-200 f/4 IS. Half the price, and still high IQ.

Plus, the tool in CS6 will allow the user to decide where to place the depth of field.

However, it's surely a tool that will improve as time goes on, and then buying that expensive (and, heavy) piece of glass might be a thing of the past.

Depth-of-field is not the only advantage of a fast lens. A lens with a maximum aperture of f/2.8 lets in twice as much light than a lens with a maximum aperture of f/4. That could mean the difference between a motion bluring 1/60 sec or a crisp 1/120 sec at the same acceptable ISO.

And I'll be amazed if CS6 can match the buttery bokeh of a 135mm F2L. I've been surprised at Photoshop before, though.

Quote:

---

Originally Posted by Sean Setters

Depth-of-field is not the only advantage of a fast lens. A lens with a maximum aperture of f/2.8 lets in twice as much light than a lens with a maximum aperture of f/4. That could mean the difference between a motion bluring 1/60 sec or a crisp 1/120 sec at the same acceptable ISO.

---

True. But, those occasions are much more limited, and probably more often than not, it's the depth of field that will be preferred (portraiture is a good example).

Plus, with newer, low noise, high ISO cameras (plus, greatly improved noise reduction software), the need for the heavier, fast lens could wane.

Certainly, the cost could be a significant deterring factor.

Quote:

---

Originally Posted by M_Six

And I'll be amazed if CS6 can match the buttery bokeh of a 135mm F2L. I've been surprised at Photoshop before, though.

---

Even if it doesn't, it won't stop the next version from imitating it more closely.

There is no substitute for getting it right in the camera to start with, there are only repairs you can do to the pictures later. I can't see how it would be a good substitute if you have to individual touch each of your photo's to get the DOF and Bokeh you want, when with a turn of a dial on your camera you can set it instantly.

If you couldn't afford a fast lens I see how a person might use this, but it still costs you. You will end up spending your time with additional processing to save the cost of the faster lens.

+1 for Sean's points. If it were just about DOF I wouldn't own the 70-200mm F/2.8L II, I would just go for the F/4 version and carry a lighter lens. It is the Wide open F/2.8 and it's ability to perform in lower light that makes me choose the F/2.8 version.

Quote:

---

Originally Posted by HDNitehawk

There is no substitute for getting it right in the camera to start with, there are only repairs you can do to the pictures later. I can't see how it would be a good substitute if you have to individual touch each of your photo's to get the DOF and Bokeh you want, when with a turn of a dial on your camera you can set it instantly.

If you couldn't afford a fast lens I see how a person might use this, but it still costs you. You will end up spending your time with additional processing to save the cost of the faster lens.

+1 for Sean's points. If it were just about DOF I wouldn't own the 70-200mm F/2.8L II, I would just go for the F/4 version and carry a lighter lens. It is the Wide open F/2.8 and it's ability to perform in lower light that makes me choose the F/2.8 version.

---

As for the first point, it's called creativity, and gives one an option that might not have existed in the situation at the time.

As for the second point, each successive program improvement makes processing easier. Besides, we all spend time tweaking images, whether globally in LR, or with actions, in PS.

Sean's point is well taken, but as time goes on, this may not be as an important factor, in light of costs (which will increase with future releases of lenses). We are so used to expecting bokeh and depth of field, and we are willing to pay for it, but if the same can be achieved with manipulation of the data (which, after all, is what we all do with every image we take, and don't take perfect pictures every time), then why not?

Maybe we'll feel upset when we realize that the $12,000 f/2.8 wasn't necessary after all, since a simple Photoshop action can give the same result?

It is the Garbage in Garbage out saying that applies. The better data you start with the better finished product you will be achieve. You can start with a picture that needs more work, but the product will be better from a picture that already has the most elements already in it.

I am not so impressed with Adobe that I believe were close to the point you are talking about. Maybe some time off in the distant future.

Quote:

---

Originally Posted by HDNitehawk

It is the Garbage in Garbage out saying that applies. The better data you start with the better finished product you will be achieve. You can start with a picture that needs more work, but the product will be better from a picture that already has the most elements already in it.

I am not so impressed with Adobe that I believe were close to the point you are talking about. Maybe some time off in the distant future.

---

That's a bit harsh, don't you think?

And, how did you do your macro photos? Was it "get it right in the camera" or did you do a photo stack with Photoshop?

Quote:

---

Originally Posted by Alan

This begs the question: who needs an expensive, fast lens?

---

Anyone who is picky enough to use a fast lens for thin DoF will be way too picky to switch to software-based thin DoF. It's like the difference between shooting real infrared and simulating it in Photoshop. Sure, the simulation might be good enough for some people, but certainly not for anyone who went to expense and effort of shooting infrared.

Quote:

---

Originally Posted by Alan

One might say that "it's not the same as what a lens can do." Perhaps? Perhaps, not.

---

No "perhaps" about it. Physically impossible. But of course, for many purposes it will be good enough. I'm sure there are already many good plugins (or standalone programs) that are superior to this new Photoshop feature for simulating thin DOF [edit: I'm guessing this because Photoshop is not really the best at anything that has a plugin or standalone alternative -- Photoshop is more of a jack of all trades]. But this one will definitely be easier to install, considering that it comes out of the box. And even without any plugins, I've simulated thin DOF with just a masked layer and gaussian blur (though I'm sure this will look more realistic than that).

Quote:

---

Originally Posted by Alan

That's a bit harsh, don't you think?

And, how did you do your macro photos? Was it "get it right in the camera" or did you do a photo stack with Photoshop?

---

No, not really harsh. It just means the better data you start with the more you have to work with to get a great finished product.

I have stacked a few, but stacking isn't something that can be done in camera. But the same saying applies, the better the pictures you start with to stack the better the product you will get with photoshop. But comparing stacking to creating a simulated DOF isn't a fair comparison. One can be created in camera, one can not. Stacking does something that isn't possible without software.

Here is a link to a video of the function you are talking about:

http://www.petapixel.com/2012/02/29/...of-field-easy/

C'mon Daniel, I'll swap you my CS4 for your 24/1.4 and we'll just call it even.

Quote:

---

Originally Posted by Steve U

C'mon Daniel, I'll swap you my CS4 for your 24/1.4 and we'll just call it even.

---

If you get his 24/1.4 will you still need your Zeiss 21mm? I have a copy of CS5 I will trade you.

Isn't that sort of what the Plug-in OnOne Focal Point 2 does already? It lets me choose the lens and the apeture ...Canon or Nikon. I can "shoot" with a Canon 85 1.2 or a Nikon 85 1.8 anytime I want ...LOL!

Quote:

---

Originally Posted by Alan

This begs the question: who needs an expensive, fast lens?

One might say that "it's not the same as what a lens can do." Perhaps? Perhaps, not.

For example, rather than buy the 70-200 2.8 II IS, one might be better off with the 70-200 f/4 IS. Half the price, and still high IQ.

Plus, the tool in CS6 will allow the user to decide where to place the depth of field.

However, it's surely a tool that will improve as time goes on, and then buying that expensive (and, heavy) piece of glass might be a thing of the past.

---

It could replace a fast lens in some occasions, however it can't in a lot of other occasions. The problem with this software is that it doesn't know where the objects on your photo are placed in real life. In the example video that Rich linked to the software does the same work as you might have done sometimes yourself with a few layers, masks and some gausian or lens blur.
This works fine for photos where pretty much everything is flat. However this doesn't work if transitions from for to background are not fluid.
For example: You take a photo from a low perspective. Person 1 stands 5 ft from the camera, person 2 stands 10 ft from the camera. You know this, however Photoshop doesn't and will create the DOF regarding what items there are in your shot. If you want to blur Person 2 say 50% and the background 100% and have that proces go smoothly, you're still assigned on doing some lovely selection-work and using multiple layers and masks to get this done.
Using a fast lens might have done the shot all at once and have been a lot easier.

As for the "flat" shots. This tool seems very easy, however just use the actions from the video below and you don't really have to upgrade to CS6:o
http://www.youtube.com/watch?v=wi90bHVPtI4

The other options is to keep following the updates on the Lytro camera. Then you can do all this with just one camera and even incamera:rolleyes: It's cheaper than a fast lens or CS6 ;)(still very expensive for a 1MP camera though...)
http://www.youtube.com/watch?v=JDyRSYGcFVM

After reading Jan’s comments about “flat” objects and watching the video I think the functions usefulness will be limited to certain pictures.

For instance this picture would not be possible to create with the DOF function:
http://farm6.staticflickr.com/5164/5...4fd4e8e9_z.jpg
Bluejay 6 by hdnitehawk01, on Flickr

I think fast lenses with good bokeh are safe for now.

Plus, there are other advantages to using high quality (usually fast) lenses; two advantages off the top of my head--better autofocus mechanisms and weather sealing (in many cases).

The Lytro camera is cool. In theory, you can control the DOF, focus plane, and the quality of the bokeh, all after the fact.

Quote:

---

Originally Posted by Jon Ruyle

The Lytro camera is cool. In theory, you can control the DOF, focus plane, and the quality of the bokeh, all after the fact.

---

Yeah it looks pretty cool. It's still a bummer that it only takes about 1MP photos(not even enough for a wallpaper :confused:) and is quite expensive at this time:( I'm not sure how fast the shutterspeed on these things is, but with an f2 aperture I guess you should be alright in quite a lot circumstances.

Quote:

---

Originally Posted by HDNitehawk

For instance this picture would not be possible to create with the DOF function:
I think fast lenses with good bokeh are safe for now.

---

To be fair, I wouldn't volunteer to try to recreate it with masks and layers either ;) I think your conclusion is right...for now(and a few years down the lne probably).

Quote:

---

Originally Posted by Sheiky

It's still a bummer that it only takes about 1MP photos(not even enough for a wallpaper :confused:)

---

Enough for web pics, but not enough for wallpaper.

The reason for the low mp count is that the thing takes 3D data, so instead of say, 1024x1024 = 1 mp for a 1mp pic, it is 1024x1024x1024 = the equivalent of 1024mp of data for a 1mp pic. That's an oversimplification, of course: I don't think it has as much resolution in the depth direction, and there are probably tricks and shortcuts. And the aspect ration isn't 1:1. But it does gather way more data than an ordinary 1mp camera.

With this camera, however, there is no reason you can't have just as creamy bokeh as any lens you care to simulate- but with the low resolution of the final pic, you may not even be able to tell :)

Lol yes you're right about that. Even though pretty much everyone will be looking at 100% views of your photos, I doubt they will congratulate your photos for their beautiful bokeh:p
If Adobe comes out with a perfected version of the Iris Blur(It should recognize depth in a shot) before Lytro upgrades their product I foresee a small problem for Lytro :rolleyes:

Quote:

---

Originally Posted by Sheiky

If Adobe comes out with a perfected version of the Iris Blur(It should recognize depth in a shot) before Lytro upgrades their product I foresee a small problem for Lytro :rolleyes:

---

Forget Lytro. If Adobe can do that, then we should have them working on a cure for cancer, cold fusion, and warp drives. ;) Unfortunately, it's just not physically possible.

Is it really?:( Isn't it possible that Adobe theoretically could recognize depth by analyzing the image for the amount of sharpness and contrast? Sort of like the Live-view AF mode? Or do I really need to do more homework?:o

Quote:

---

Originally Posted by Sheiky

Isn't it possible that Adobe theoretically could recognize depth by analyzing the image for the amount of sharpness and contrast?

---

In certain contrived circumstances, yes. But not for any (or even most) photos. The reason is that there are too many possible inputs (scenes) for the given output (a 2D image). For example, consider a test where you have two perfectly identical photos, but one was made using a large subject far away and the other was made using a small subject that is closer. The result is perfectly identical, so there is no possible way for Adobe to know which one is distant and which is close.

As another example, consider shooting through a chain-link fence. In the narrow f-number version, the fence obscures many details -- there is no way photoshop can get those back by blurring the fence with simulated thin DoF. If a wide f-number lens was used instead of software, the fence would be blurred into nothing and all the details would be visible on the subject (though overall contrast would be reduced, of course).

Hope that helps explain why it's impossible.

Daniel, don't let facts get in the way of a good theory. Sheesh.

Hehe!(I had already started on writing a never seen before algorithm for it: Mission Impossible X):) But thanks Daniel, that makes sense.

I'm going to stick with fast lenses over a new trick in photoshop for two reasons:

1. The better low light usability
2. I like shiny toys with giant front elements.

There's no way photoshop or any other post production software can replace the performace of a high quality, wide aperture lens. I'll gut my 70-200mm f/2.8L IS II and use it as a water jug if Adobe can prove me otherwise.