Canon 70D - an idea- digitally enhanced bokeh- Magic Lantern or Firmware?

[neoday2]

Good morning- this is a thought I had this morning on the way in to work. With the 70D's dual pixel design, would it be possible to "digitally enhance" the bokeh of images? Could you load bokeh profiles since the camera knows what is in focus and what is not? Someone- Magic Lantern (when available), Canon, or even Adobe Lightroom could create bokeh profiles for the parts of the image that are not "in focus" based on what the sensor is reporting/sharpness.

Or does something like this already exist? This way, you could get F1.4 bokeh on almost any lens at any F stop.

Your thoughts?

 

[Mt Spokane Photography]

Just what specifically would you enhance? Bokeh is not a well defined term and means different things to different people. It is not something that can be measured, its subjective. First, a method of measuring it would be needed, and so far, no one has one.

 

[neoday2]

By "enhance" I mean to increase the power of the out of focus capabilities. Larger apertures (assuming same focal length) produce more of the "blurr". Each lens has different bokeh characteristics- number of aperture blades, elements, etc impact how bokeh is rendered. You could create bokeh profiles that add additional softness (read out of focus/depth perception) to the image.

Think about the bokeh difference when taking a picture with a 50mm lens at F8 vs F1.2 and a subject that is 10' away. The images would be substantially different.

Now, if you wanted to get a similar bokeh to the F1.2 but your lens is only capable of shooting F8 that would be impossible (without lots of time in photoshop). But since the sensor can detect the phase, can it then enhance the image with the bokeh profile you select- much like choosing Portrait, Landscape or any of the other image enhancements.

It might end up looking like an Instagram photo but I gotta believe it would be better because of the additional phase information provided by the sensor and focus point?

 

[David_in_Seattle]

Good morning- this is a thought I had this morning on the way in to work. With the 70D's dual pixel design, would it be possible to "digitally enhance" the bokeh of images? Could you load bokeh profiles since the camera knows what is in focus and what is not? Someone- Magic Lantern (when available), Canon, or even Adobe Lightroom could create bokeh profiles for the parts of the image that are not "in focus" based on what the sensor is reporting/sharpness.

Or does something like this already exist? This way, you could get F1.4 bokeh on almost any lens at any F stop.

Your thoughts?

That's not quite possible as bokeh is formed with the correlation of sensor size, aperture, and distance to subject. Now, you can digitally add a similar affect via photoshop, but there's no firmware that can match PS's capabilities straight from camera. Though time it would take to do so would not be cost productive, plus there are other added benefits with using a wide aperture lens, namely the ability to shoot in low light.

 

[neoday2]

I guess I'll have to play around some more with it.

I understand there are benefits to shooting @f1.4 vs f8. But this weekend, when running around taking pictures of the egg hunt at f1.4 (Canon 50mm f1.4) in daylight with shutter speeds of 1/1,000-1/4,000 I could have been shooting at F8 or higher and been able to select the amount of bokeh in post.

Since bokeh is the correlation of sensor size, aperture and distance to subject- we should know all these things based on the sensor readout. Thus be able to mathematically represent the impact of these individual elements on the light that reaches the sensor. Knowing this information would then also allow us to do additional calculations on the information gathered and create new bokeh profiles.

Like here:
http://www.scratchapixel.com/lessons/2d-image-processing/simple-image-manipulations/example-1-simulating-the-bokeh-effect-in-2d/

Anyway, just an idea. Thank you all for the input!

 

[David_in_Seattle]

I guess I'll have to play around some more with it.

I understand there are benefits to shooting @f1.4 vs f8. But this weekend, when running around taking pictures of the egg hunt at f1.4 (Canon 50mm f1.4) in daylight with shutter speeds of 1/1,000-1/4,000 I could have been shooting at F8 or higher and been able to select the amount of bokeh in post.

Since bokeh is the correlation of sensor size, aperture and distance to subject- we should know all these things based on the sensor readout. Thus be able to mathematically represent the impact of these individual elements on the light that reaches the sensor. Knowing this information would then also allow us to do additional calculations on the information gathered and create new bokeh profiles.

Like here:
http://www.scratchapixel.com/lessons/2d-image-processing/simple-image-manipulations/example-1-simulating-the-bokeh-effect-in-2d/

Anyway, just an idea. Thank you all for the input!

Going from f1.4 bokeh to f8 bokeh is impossible. You could add background blur to an f8 image's bokeh to make it simulate f1.4, but the time it would take to do this isn't cost effective. In addition, DSLR manufacturers will never incorporate such firmware into their cameras because it would negate the reason for customers to buy their higher end lenses. Plus, simulated Bokeh is still nowhere near what a physical lens and camera can create. The aperture blades also have an impact on the visual shape and quality of the Bokeh.

Lastly, there are cameras that can do this, but the technology is still in it's infancy. Instead of a 2D sensor it incorporates a 3D sensor. Lytro is a consumer brand that currently sells cameras with this technology, though it's more gimicky as the image quality doesn't come close to what a DSLR or even a compact camera can provide. Here's their website: https://www.lytro.com/go/holiday-bundles/?gclid=CKj2z9278r0CFQEepQodeisAlw

 

[Marsu42]

Since bokeh is the correlation of sensor size, aperture and distance to subject- we should know all these things based on the sensor readout.

Nope, there's your problem - the camera only knows the distance to the focused object from the af system, everything else is 2d. To any fancy calculations based on distance like bokeh, you'd need to know real 3d information and generate a depth map - probably then you'd be able to simulate different aperture sizes in post, most likely in the small->large direction.