Jump to content

Line Art from a color image


Recommended Posts

Hi. I'm a newbie with this program, so forgive me if this is a dumb question. I am busy on a new project and want to convert color images to line-art that would be suitable for use in a coloring book.

I've done as much as I think I can, but the results are not acceptable to me.

I desaturate, do the ink-sketch effect and adjust brightness and contrast. This leaves too much 'clutter'.

Am I missing something, or is there a decent technique that works well?

 

post-141262-0-20165300-1430991113_thumb.

post-141262-0-99720200-1430991128_thumb.

Link to comment
Share on other sites

Link to comment
Share on other sites

I doubt those plugins will really help Msomi do what s/he wants to do. What Msomi wants, as I understand it, is a method that will produce a "clean" line drawing from a photograph; that is, one with well-defined black lines on a white background. I wish I had a good suggestion on how to do that, but I don't know of a plugin or a series of operations that can do that; especially with an image like the wild cat, which lacks sharp color boundaries. Crayola has a website that supposedly turns photos into cartoons (it requires buying Crayola products to register). The "examples" they show make it look like it would do a great job, but it's actually terrible. I think it's a pretty difficult problem.

 

Neil Cassdy has a color segmentation plugin that might possibly be helpful for something like this. The images could be color segmented, then the color regions outlined. I don't know how well that would work.

Link to comment
Share on other sites

  • 2 weeks later...

In order to shorten the wait, I'll give the basic steps I followed for the picture of the cheetah in the first post. You will need the Apply Texture plugin.

  1. Set the picture's colors to black and white and duplicate it.
  2. On the topmost layer, run Gaussian Blur with an amount of around 7, then invert colors and set opacity to around 128. Then merge both layers. This step is intended to produce a High Pass filtering. I also tried with the High Pass filter plugin but didn't like the results. Perhaps it's just a matter of personal taste or a matter of combination of values I couldn't find, so I just stuck with the method described.
  3. Duplicate, set upper layer's blend mode to Color Dodge, and merge. This creates the sketchy strokes on the white background.
  4. Now duplicate three times (there should be four layers in total). Set all blend modes to Multiply and flatten. This is to darken the sketch, therefore making it more defined.
  5. Create a new layer, fill it with white, and set its blend mode to Color Burn.
  6. Copy the original image of the cheetah to the clipboard.
  7. Make the layer created in step 5 the active layer and go to Effects > Object > Apply Texture (click where it reads Get from Clipboard and keep brightness as it is). Set colors to black and white, run Curves+ so as to emphasize the dark shades in the cheetah's mouth and right side, according to the original photo, and finally run Auto-level. The purpose of this step is to preserve some of the dark and light shades of the original image so as to make the sketch more defined and more close to the original photo, or at least that's how I feel it looks better. I also tried other methods to define the dark and light shades, but this is the approach whose results I liked best.
  8. After flattening, you can go on tweaking details as you wish, like polishing shades with the paintbrush or something like that.

 

These steps don't always have the same results on different pictures and here's where you need to experiment a lot. Sometimes they don't even produce exactly the same results on the same picture when done more than once. For example, you can experiment with different blend modes and opacity settings in steps 2 and 3, the Gaussian Blur value in step 2 may be set to something other, and so on.

 

Some pictures contain noise that needs to be dealt with between steps, for example by using the Surface Blur or the Reduce Noise plugins or sometimes a combination of both at different settings.

 

It's a matter of trial and error and checking what looks best to one's personal taste, so good luck!

 

By the way, I'm not sure if all these discoveries are worthy of a more detailed tutorial. I've been trying to find tutes on how to turn a photo into a sketch both on the forums and all over the web but haven't been really successful. The only tute I've found to be satisfactory is one for GIMP, which is the one method I've been trying to simplify and adapt into Paint.NET with varying degrees of success.

  • Upvote 1
Link to comment
Share on other sites

One thing I definitely don't care for, though I can't offer any alternatives right now, is converting the image to back and white before finding the edges. I throws away valuable information. Suppose you have a clear blue and green edge, but the shades convert to about the same intensity. You lose an edge that was in the original picture.

Link to comment
Share on other sites

Makes sense, although let's remember that sketches are not necessarily supposed to have a lot of detail (unless you want to produce a really detailed sketch). However, it's a point worth checking out. I'll go on testing variations of the method excluding the black and white conversion and see if I can find better approaches for detecting edges. I have to say all this edge detection thing is quite a tricky topic. It seems there's no surefire method that could be applied to almost any picture.

Link to comment
Share on other sites

I tried your procedure, but without converting to back and white until the final step, and the results were very similar to yours. I'm not sure what would happen if I tried it with an image with more color variation, but I'll try soon. I think this wild cat image is quite challenging for any edge-detection method, since the edges aren't very definite. As I mentioned, your method works better than any of the various ideas I tried, particularly around the cat's left eye, which often showed up as a rather indistinct blob.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...