5D4 and DPP

[RobertTheFat]

In B&H's online presentation yesterday on the 5D Mark IV, the Canon representatives were at great pains to point out that some of the camera's esoteric new features depend on post-processing via Digital Photo Professional. Asked whether those features will be supported in Lightroom, one of them declined to predict whether or when Adobe, or any other third-party software vendor, would step up to the challenge.

I think that leaves Canon in a bit of a bind. Version 4 of DPP is, on paper, a very capable photo editor with a wealth of useful features. (Version 3 lags far behind and is no longer a serious player.) But its implementation is atrocious. It's buggy and painfully slow, and its GUI contains actions that are ambiguously defined (think "Save") or are unnecessarily (and confusingly) different from their implementations in previous editions of DPP. DPP's advocates (I'm one of them) have dutifully looked the other way for more than two years, hoping, against mounting evidence, that Canon would get its act together and straighten the product out. Clearly this hasn't happened, and the most recent release of DPP 4 was more than five months ago. Lightroom's users are openly contemptuous of DPP; and if they have to start using it in order to take full advantage of the 5D4, their reaction is going to be interesting (and probably entertaining), to say the least.

I guess the best that we can hope for is that the release of DPP 4 that introduces support of the new features will also clear up some or all of the existing problems. I wish I could say I thought that such an outcome were likely.

[Waddizzle]

I believe they also announced a new release of DPP4, too.  I suspect most of the issues with DPP4 could be related to the fact that it seems to be quite the memory hog.  I upgraded a couple of laptops with more RAM[ from 4 to 8GB[, and DPP4 seems to run a little better.  I suspect that it would run best with 16-32GB of RAM, though.

[TTMartin]

I'm sure there will be companies that develop software to exploit the dual pixel RAW file. 

While Canon downplays the Lytro like capabilities of the files, the fact remains that every pair of dual pixels in an image will contain distance information. So an image taken with a small aperture and a large depth of field, can be computer manipulated to create an artificial plane of focus, that could move for and aft through the image. All pixels at a given distance would remain as captured, as the pixels move further forward or back from that artificial plane of focus they could be given an increasing amount of gaussian blur, thus mimicking the capabilities of a Lytro camera.

[ScottyP]

The Lytro effect would be nice for sure. I do assume Adobe can make everything work with Lightroom, because I also doubt the pro community believes DPP is ready for Prime Time alone if Adobe cannot work on the new RAW in LR or PS.

[kvbarkley]

http://www.learn.usa.canon.com/resources/articles/2016/eos-5d-markiv/eos5d-markiv-dual-pixel-raw.sht...

I think any "Lytro effect" will be modest at best since all the pixels are in one plane of focus and the dual pixels are not far enough apart to get much information out of them.

[TTMartin]

---

@kvbarkley wrote:

http://www.learn.usa.canon.com/resources/articles/2016/eos-5d-markiv/eos5d-markiv-dual-pixel-raw.sht...

I think any "Lytro effect" will be modest at best since all the pixels are in one plane of focus and the dual pixels are not far enough apart to get much information out of them.

---

The dual-pixels get enough distance information to auto focus in live view. So I'm not sure how much more information you think it would need. The key to what I said was that you take a large depth of field photo, and simulate out of focus areas. This is significantly different than being a Lytro camera. But, it could give the same effect.

[kvbarkley]

Having the information is not enough. You need a bunch of areas set at various focus distances, which is what the Lytro does. In this case the lens still has one focus point. You might be able to tell just how far out of focus any point is, but it won't let you magically bring it into focus. You can try to de-convolve it, but I don't think it will work, without knowing a *lot* about the lens bokeh.

[Waddizzle]

---

@kvbarkley wrote:

Having the information is not enough. You need a bunch of areas set at various focus distances, which is what the Lytro does. In this case the lens still has one focus point. You might be able to tell just how far out of focus any point is, but it won't let you magically bring it into focus. You can try to de-convolve it, but I don't think it will work, without knowing a *lot* about the lens bokeh.

---

I don't think one should expect anything remotely close to what Lytro does.

" Tied into the Dual Pixel sensor architecture is a brand new feature for Canon: Dual Pixel RAW, which allows for fine tuning of certain image parameters, including bokeh shift and focus micro adjustment, after the image is taken. "

"Dual Pixel RAW

First implemented here on the 5D Mark IV, Dual Pixel RAW allows photographers to record all the information the sensor's unique pixel architecture can deliver, providing a way for fine adjustments to be made after the image is taken. These adjustments include focus fine tuning that can help you ever-so-slightly adjust the focus point to bring out extra details, shift bokeh around for improved composition, and reduce the impact of ghosting on your photos. This requires the use of Canon's Digital Photo Professional 4.5 software."

That's taken from the description literature of the 5D4.  I get the impression that any shift will be modest at best, with perhaps just enough control range to allow you to fine tune the focus AFTER a picture has been taken.

[kvbarkley]

Go to my link from the DLC and you can see just how subtle it is. But getting back to TTMartin's point, decreasing DOF would be much easier than increasing it!

[kvbarkley]

Hm, we already have exposure bracketing, why not add focus bracketing?