For some reason when I use Santiag it seems to make the image a bit brighter and or change colors a tad bit.
It also seems to be adding artifacts over the video, they look greenish or yellowish and appear over the colors. I guess if you look hard enough, they seem to already be there slightly and are hard to see, but using Santiag makes them very visible or much more noticeable.
here are a couple screenshots to show what I am talking about.
(With Santiag x2) Used it x2 so its easier to see
Look at the guys Blue Jacket for example. it looks like a mess of colors or artifacts with Santiag being used and it seems to be making white lines along the black lines more visible as well.
I circled the artifacts on the jacket and or whiter lines against black lines that I'm referring to in this image.
If you rapidly flip between the two, you can see the color and or brightness differences when Santiag is being used.
I have tried messing with all the parameters and the only parameter that fixes it is if I change STRH and STRV from the default 1 to 0, but if I do that, it doesn't antialias it any at all.
Is there anything I can do about this mess with the colors on the guys jacket?? I guess if you look hard enough, they seem to already be there slightly, but using Santiag makes them very visible or much more noticable.
Is this a colorspace issue?
+ Reply to Thread
Results 1 to 7 of 7
Last edited by killerteengohan; 18th Nov 2015 at 13:45.
Still not having much luck with this after a couple hours of trying.
Anyone figured it out yet?
Is there a good alternative to Santiag that isn't too damaging?
Santiag uses a sharpening resizer. That is accentuating/creating the oversharpening halos.
Trying to blur it after using santiag doesn't really seem to remove the halo's it made more visible in it.
I tried using dehalo filter afterwards and it kinda screws up the antialiasing and makes lines a bit jagged again.
Aside from the Halo's though, since they don't bother me as much, the thing really bothering me is being able to notice those color screw up or artifacts like on the blue jacket alot more after using it. Got any ideas about that one?
I thought maybe mctemporaldenoise might reduce it a little. I haven't tried it yet.
Starting with your source image I applied dehalo_alpha(rx=2, ry=2) after Santiag() -- it removed the halos pretty well and didn't increase the aliasing or blur the image. It doesn't help with the accentuated noise in the grey jacket. But you can work around that by building an edge mask and replacing large flat areas with the original image.
before = ImageSource("before.png").ConvertToYV12() after = before.Santiag().Dehalo_alpha(rx=2, ry=2) edges = after.mt_edge().mt_expand().mt_expand().GreyScale().Blur(1.0).Blur(1.0) final = Overlay(before, after, mask=edges) Interleave(before, after, final)
Wow I didn't know I could do a filtering to a video and then put some of the original image back on over top of it untouched. That sounds so useful!
I am going to have to play with that some and see if I can figure out how to fully use it.
I will give it a try tonight.
mt_edge() is used to build a map of edges in the frame, black where there are flat areas, white where there are sharp edges. Overlay then uses that map to determine which parts of the final frame are taken from the first (before) frame and which parts come from the second (after) frame. Where the map is black pixels will be taken from the first source. Where it's white pixels will be taken from the second. Where the map is some shade of gray pixels will be taken as a weighted average of the two sources.