Jump to content

BornInPastCentury

Members
  • Content Count

    15
  • Joined

  • Last visited

About BornInPastCentury

  • Rank
    Newbie

Recent Profile Visitors

403 profile views
  1. Hi. I've encountered error developing image with haze remove filter. Here is the original image crop: : And here is "dehazed" one: Dehazing was performed on a uncropped image. As it can be noticed, the palm makes "depth recognition" algorithm to fail. And here's smaller version of uncropped image: and how "dehaze treated it: As it can be seen, this time the palm was not a problem. Any ideas on how to make it work properly? Best regards. BiPC
  2. There are numerous publications on the depth estimation subject. Many uses convolutional neural networks, which is machine learning class algorithm rather than AI. Whatever "engine" uses our Affinity, I think that utilization of KNN is not impossible.
  3. Hi. Nice and very informative (with help of little reverse engineering ). However, when this macro is executed for image with little to no haze, the resultant mask seems to not show the depth... Result may indicate that the macro (and dehaze filter itself) estimates amount of haze rather than the scene depth.
  4. Hello. I think that it might be a cool feature of the "haze removal" to create a layer's mask, instead of modifying the layer content. I assume that this plugin does scene depth estimation. I so, it should be possible to generate mask with levels corresponding to the estimation results. This mask could be use with levels, curves and other tools (or other bitmap layers). Regards. BitPC
  5. I can confirm this issue. RGB file looks the same, CMYK file looks different (Affinity displays more saturated image).
  6. Checking this filter I've done an assumption, that median blur is implementation of one of a "classical" 2-d median filter algorithms. My first observation: Filter's slider shows values with one decimal position accuracy. What does "1,7 px" radius mean? It looks like program truncates (or performs round operation), but here is little confusion. My second observation: Preview of the resultant image with scale different than 100% is little weird. When this scale is eg. 75% then median with "0px" diameter shows some changes in the image preview (aka "after"). Considering my assumpti
  7. Still requires to rasterize the image if I want to use inpainting brush to fill corners of a image straightened with the crop tool ("straighten" button of the crop tool).
  8. Thanks, there is echo but with big delay :) Rasterize helped, inpaint semes to work o'k now. Not sure what rasterize does for layer being marked as "pixel", but it cured problem. I hope it will be fixed in the release version.
  9. "Portrait - format" image (From Canon EOS REBEL T5) opens as "landscape" in NIK Colorefex. error is present since first release of beta.
  10. Hello. I've noticed, that inpaint brush works insanely slow. Here is scenario: Image (Eos rebel 5, "portrait" format) has "tilted" horizon. Image is straightened using "straighten" function of crop tool. As a result I have image with "missing" content in corners. After applying inpaint brush to "fill" them i got three possible results: - filled or partially filled (with some "empty" gaps) areas with "proper" content, - no changes at all. Sometimes this action takes 2...3 minutes on my notebook (I7 2630, 16G ram, Nvidia GT540), sometimes it is reasonably fast (10 seconds). How it works in
  11. I've noticed following problems: - "rotate 90 clockwise" has two entries in Document menu, while one is for counterclockwise, - I straightened image (just few degrees, using straightened option in the crop tool) and next executed "merge down" from layer palette. Gray chessboard showing background remained in the "triangles" in the image corners. - "portrait" image (jpg) from EOS rebel T5 opens in NIK viveza and colorefex as "landscape" (not tried other filters). Removing EXIF in Irfan fixed problem. - straightened image has some triangles with "chessboard" artifacts after merging down. Im
  12. Hi. Are AOpponent and BOpponent intentionally named this way? Regards.
×
×
  • Create New...

Important Information

Please note there is currently a delay in replying to some post. See pinned thread in the Questions forum. These are the Terms of Use you will be asked to agree to if you join the forum. | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.