Jump to content
Lojza

[Fixed] Depth of Field filters provide bad results (1.6.5 beta 5)

Recommended Posts

Both destructive and live Depth of Field filters, in Tilt Shift mode, render visible errors in the "smoothness" of the blur gradient. See the screenshot. These non-fluent blur levels emerge all along the axis of the Tilt Shift configuration. This makes the filters vastly unusable for sufficient results.

screenshot_0785.thumb.png.8eaf6fce8b76ede9a6f570a8f6e9769e.png

Tested on Sierra 10.12.6, Apple Cinema Display attached to MacBook Pro, display mehod - Metal.

Thanks for looking into it.


UX/UI designer, IT analyst & consultant, Business Architect at Cool Ticket (www.coolticket.co).

MacBook Pro 13'' Early 2015, 3,1 GHz Intel Core i7, 16 GB 1867 MHz DDR3, Apple Thunderbolt Display 27'' (2560 x 1440).

Share this post


Link to post
Share on other sites

Hi Lojza,

 

I've just tried this myself on High Sierra using Metal and I couldn't see any issues. Would you be able to attach a image using a Depth of Field Live filter that shows the issue please?

 

Thanks.

Share this post


Link to post
Share on other sites

I had the same issue with Mac pro 2013 and metal when I made focus stacking with 30 (sony) raw files and then tried to add Live Depth of Field layer. It was so slow anyway that I needed to export it to png before continuing editing and didn't try that live filter with it again. I can try it again if needed :)

Share this post


Link to post
Share on other sites
On 10/3/2017 at 0:50 PM, Sean P said:

Hi Lojza,

 

I've just tried this myself on High Sierra using Metal and I couldn't see any issues. Would you be able to attach a image using a Depth of Field Live filter that shows the issue please?

 

Thanks.

I'll send you a PM with non-public link. Thanks.


UX/UI designer, IT analyst & consultant, Business Architect at Cool Ticket (www.coolticket.co).

MacBook Pro 13'' Early 2015, 3,1 GHz Intel Core i7, 16 GB 1867 MHz DDR3, Apple Thunderbolt Display 27'' (2560 x 1440).

Share this post


Link to post
Share on other sites

The strength of blur is stepped across the image, rather than the infinitely smooth transition of a physical lens. The result is bands of blur akin to the colour banding in gradients that can be seen in 8 bpc images.

The blur banding can be particularly noticeable in regions containing high contrast because the filter (as do all Affinity blur filters, and those of most 2D graphics apps) makes the classic mistake of blending a non-linear image's gamma-adjusted pixel values instead of blending the pixels in linear colour space.

 

If you switch colour mode to 32 bpc linear, do the blur there and bake it in (i.e. don't retain a live filter blur), and then switch back to 16 (or 8) bpc mode, there will still be banded blur but it should be less noticeable because the new pixel values will have been correctly calculated in linear space.

 

Share this post


Link to post
Share on other sites
2 hours ago, Lojza said:

I'll send you a PM with non-public link. Thanks.

Thanks for the file - I've reproduced it with that. I've tried switching to both 16 and 32bit and still got the banding. So I've passed it onto development to see what they say!

Share this post


Link to post
Share on other sites
1 hour ago, Sean P said:

Thanks for the file - I've reproduced it with that. I've tried switching to both 16 and 32bit and still got the banding. So I've passed it onto development to see what they say!

 

 

Yes, as I said, 32 bpc does not prevent the banding. 32 bpc should make the banding somewhat less conspicuous in regions of high contrast for the reason stated.

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×