1

Adding Ashwarya Rai's face to the Mona Lisa painting## Heading ##

Hello

GOAL: TO MAKE SKINTONE MATCH IN FACE AND NECK

Newbie. Just started learning Photoshop.

I have a practice exercise to modify the Mona Lisa in whatever way I want (just to practice.) I decided to give her a celebrity's face. I am using this awesome Photoshop Training Channel Tutorial. I am adding Ashwarya Rai's face to the Mona Lisa painting.

First, I

  1. Increased the vibrance and saturation of the original image which looked too dull (perhaps a mistake)
  2. Then I copied in Ash's face after a lasso selection.
  3. Resize, move
  4. Ctrl T to transform and warp, using right eye as my reference
  5. Auto blend (Seamless fill and Content Aware checkboxes and Panorama selected)
  6. Layer mask to clean up edges

I notice (channels) the Mona Lisa was mainly reds with yellow, almost no blue Ash's face has more blue

I've tried fiddling with the image adjustments and layer adjustments. I want her face to blend in.

*What should I do better?

  • I can do over fixing the original image first
  • Edit the face source and target Mona Lisa before swapping What should I do to make the faces match? What adjustments should I make this time round?* Thanks!
7
  • There's one odd thing about the Mona Lisa that your overlay makes even weirder… she already looks like her face is turned more 'towards camera' than the rest of her head. You overlay emphasises that still further. In more practical terms, get rid of your initial 'colour enhancements' entirely. It makes it look like a cartoon. – Tetsujin Nov 19 '20 at 11:53
  • #1 aspect needed to blend things well.. matching angles of source images. Clearly the faces in the source images are not the same. You need to find a photo which matches the angle of the painting. You can't "fake", "force", or "correct" this aspect.. it all comes down to source material. After that.. color blending is comparatively easy. – Scott Nov 19 '20 at 12:31
  • @Scott, with the new Neural Filters in Photoshop you can actually turn the head slightly with a slider. Check this video. It's still very experimental and I hate the idea and all, but it is somewhat possible. 😀 – Wolff Nov 19 '20 at 13:14
  • 2
    @Wolff - Wow, that's eerie. How to turn a human into a bad CGI figure in 3 easy steps ;) – Tetsujin Nov 19 '20 at 13:31
  • That tutorial is not good at all. Matching the eyes is a total mistake. 1. The face is not just the eyes and mouth, but the overall shape and proportion of the head. 2. It is supposed to "match" the colors but as you can see, it did not, It is not about colors. It is about everything. – Rafael Nov 19 '20 at 15:18
1

What should I do better?

One thing: Observe

And that is starting with your choice of starting material. Please, do not use a direct flash photo of a person... that is the ugliest light anyone can have.

Here is a search on the actress:

enter image description here

Which image has the best potential to match the intended usage?

A. Matching angle

B. Matching light

Gosh, even the hairstyle matches!

enter image description here

Answer: Observe


Now, to answer the 2 questions about color.

1. Reducing the yellow

That color is probably now an intrinsic part of the look of the Monalisa. Yes, they are not the original colors, it is hundreds of years old... but the exercise probably intends to make it look like a renaissance painting, so, keep it to some extent.

Simply use curves and... observe.

Yellow is the opposite of blue, and as the painting is dark, I do not want to reduce the red and green, so I will rise a bit the channel blue.

enter image description here

Done. The image less yellow but still looks old and natural.


2. Match the photo to the overall tone of the painting

Well, let's try just the opposite, reduce the blue instead of rising it.

enter image description here

Observe that I did not use a curve, just a straight line, that is because I want to reduce the white from the eyes. The painting does not have white for the eyes.

I still need to adjust some things. Reducing the green so the image looks a bit more magenta.

enter image description here

And rising the dark point. I need to raise the red a bit more to give it a brownish tone.

enter image description here


I adjusted a bit more the curves on both layers.

I am doing very lazy work. But even without doing any blending, masking nor scaling, the project looks promising... weird, but promising.

enter image description here

The point, again, is NOT a recipe, the only real recipe is: Observe, and based on that, make decisions...


An additional note:

Do not match a face based on the eyes and mouth position. I was tempted to use transparency to "match" but that is a mistake.

The only way to match the scale and position is to observe the image and feel it right.

1
  • Hi @Rafael Thank you for all the tips! Exactly what I was hoping for. I will redo this project with your feedback in mind, and observe more. – codingisfun976 Nov 22 '20 at 15:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.