affinity photo Completing my fanfiction composite - Working with Inpainting

[SolidSnake2003]

I have tried the next step to completing my composite image by trying to hide the original stuffed animal from the original image. Here is what I have with just taking the original image, and used the inpainting of the fill option in the Edit Menu Then I masked in what parts I wanted visible.

[SolidSnake2003]

Thanks in advance for any help provided

[SolidSnake2003]

Just bringing this back up to the front.

[SolidSnake2003]

How did I do on using the inpainting feature of Affinity?

[SolidSnake2003]

Just bringing this back up to see what everyone thinks of how I did on my first use of the inpainting feature.

[firstdefence]

The inpainting looks good. 

[TrentL]

On 7/28/2023 at 6:41 AM, SolidSnake2003 said:

I have tried the next step to completing my composite image by trying to hide the original stuffed animal from the original image. Here is what I have with just taking the original image, and used the inpainting of the fill option in the Edit Menu Then I masked in what parts I wanted visible.

I don't know what your original image looked like, but the teddy bear looks really good. The only thing that looks a little out of place to me is the light on the kid's face. 

[SolidSnake2003]

2 hours ago, TrentL said:

I don't know what your original image looked like, but the teddy bear looks really good. The only thing that looks a little out of place to me is the light on the kid's face. 

Here is the original unaltered image:

It is a screenshot from an episode of the latest season of The Good Doctor. Here is a screenshot of the original face:

 

This is a screenshot from the movie, The Prodigy that came out in 2019. I work with screencaps mostly cause I can't afford stock photo prices.

[SolidSnake2003]

I tried to match the color as best as I could using color balancing, selective coloring and curves. I also did all of it using just a mouse as I do not own a drawing tablet, doubt I would good with one. I cannot draw. I also used check layers to try to match the coloring and luminosity as much as possible.

[SolidSnake2003]

On 7/31/2023 at 8:03 PM, TrentL said:

I don't know what your original image looked like, but the teddy bear looks really good. The only thing that looks a little out of place to me is the light on the kid's face. 

I have attached the original photos in a post above, do you have any advice on how to improve the color matching?

[SolidSnake2003]

Is there any more advice on how to finish this?

[TrentL]

45 minutes ago, SolidSnake2003 said:

Is there any more advice on how to finish this?

I think it looks pretty solid. My advice for the light matching would be just to look at it visually. Maybe just add a little bit of soft light so the left side (on our left) isn't so dark. I think even with a mouse, a soft brush and the right blend mode could do the trick.  The hard part is that in the screenshot of the original face, the lighting is much different than the face you are replacing. But overall I really like it. What is your ultimate goal with this piece?

[SolidSnake2003]

55 minutes ago, TrentL said:

I think it looks pretty solid. My advice for the light matching would be just to look at it visually. Maybe just add a little bit of soft light so the left side (on our left) isn't so dark. I think even with a mouse, a soft brush and the right blend mode could do the trick.  The hard part is that in the screenshot of the original face, the lighting is much different than the face you are replacing. But overall I really like it. What is your ultimate goal with this piece?

Thanks for the advice, My goal with this piece is that it is an illustration for for a fanfiction scene. I don't plan on illustrating every scene, but this is one I found  a scene I could work with. This is a scene from an IT fanfiction where Georgie actually survives Pennywise's attack, and he's in the hospital after getting stitches for cuts in his arm. He's sleeping after being given a sedative to help him calm down.

[SolidSnake2003]

@TrentL Here is an updated look at the image, I have adjusted the shadow on the face, adjusted the colors to match better and adjusted the face a little to match the original better:

[TrentL]

1 hour ago, SolidSnake2003 said:

@TrentL Here is an updated look at the image, I have adjusted the shadow on the face, adjusted the colors to match better and adjusted the face a little to match the original better:

Nice work, I really like what you did with the lighting 😎

[iconoclast]

For me, the lighting looks a bit confusing. Bedause there is a light on the face of the child, that comes from the front rigth, while the face of the mother - and even her arm and the teddy - has a shadow on that side. The same with the linen behind the child. And the childs face has a very dark shadow on the left side (from our perspective). You inserted the child there, right? If nobody perceives it, it might be OK.

[SolidSnake2003]

8 hours ago, iconoclast said:

For me, the lighting looks a bit confusing. Bedause there is a light on the face of the child, that comes from the front rigth, while the face of the mother - and even her arm and the teddy - has a shadow on that side. The same with the linen behind the child. And the childs face has a very dark shadow on the left side (from our perspective). You inserted the child there, right? If nobody perceives it, it might be OK.

@iconoclast I didn't insert the child into the image, I just replaced the face. Here is the original unaltered base image: https://i.imgur.com/qnGW5By.png It is a screenshot from an episode of The Good Doctor. This is the original image used for the face replacement: https://i.imgur.com/wQHrJI4.jpg this is a screenshot from the movie The Prodigy from 2019. These are the best images I can work with. The closest other image I can use is this one, but it is too small and low quality: https://i.imgur.com/7ldZFf1.jpg 

[SolidSnake2003]

@iconoclast I am trying to match the lighting of the original as best as I can. I do no have a wacom tablet or any other kind of tablet. I am doing all of my work with a mouse and using low flow and opacity settings.

[iconoclast]

But what was on that image before you replaced the boy? And how was it lighted?

The lighting is one of the most important things in image compositing. If the lighting of the images you assemble together, doesn't match, it will always look strange, even the viewers often can't exactly put the finger on what ist wrong. It's somehow similar to assemble images with different perspectives. That would look strange too. It is often difficult to find the right images you need for that. And it is even more difficult to adapt images with a wrong lighting angle to another one.

As I heard some time ago, Artificial Intelligence can do such things. Photoshop or a plugin for Photoshop can eventually do it. But I'm not sure about that. Think of, that the whole lights and shadows of the image fragment you insert, must be replaced, so that would mean, that you would need to overpaint the main parts of the image. You must be very experienced in perspective and lighting for that, to create a persuasive illusion. And in painting too. For that, in fact a graphics tablet would be helpful. But it will not do the whole job.

In my experience, the best way to do such things ist to find images that fit to each other.

[SolidSnake2003]

2 minutes ago, iconoclast said:

But what was on that image before you replaced the boy? And how was it lighted?

The lighting is one of the most important things in image compositing. If the lighting of the images you assemble together, doesn't match, it will always look strange, even the viewers often can't exactly put the finger on what ist wrong. It's somehow similar to assemble images with different perspectives. That would look strange too. It is often difficult to find the right images you need for that. And it is even more difficult to adapt images with a wrong lighting angle to another one.

As I heard some time ago, Artificial Intelligence can do such things. Photoshop or a plugin for Photoshop can eventually do it. But I'm not sure about that. Think of, that the whole lights and shadows of the image fragment you insert, must be replaced, so that would mean, that you would need to overpaint the main parts of the image. You must be very experienced in perspective and lighting for that, to create a persuasive illusion. And in painting too. For that, in fact a graphics tablet would be helpful. But it will not do the whole job.

In my experience, the best way to do such things ist to find images that fit to each other.

Here is the base image with no edits:

[iconoclast]

OK, if you look at the nose of the boy on the base image, you can see, that it has a shadow on the right side (from our point of view), corresponding to the lighting on the neck of the mother. The lighting on the face of the boy comes from the front left side too. The lighting on the face you inserted comes from the front right side. And it has a harder contrast. I would at least lower the contrast.

[SolidSnake2003]

11 minutes ago, iconoclast said:

OK, if you look at the nose of the boy on the base image, you can see, that it has a shadow on the right side (from our point of view), corresponding to the lighting on the neck of the mother. The lighting on the face of the boy comes from the front left side too. The lighting on the face you inserted comes from the front right side. And it has a harder contrast. I would at least lower the contrast.

Ok here I lowered the contrast with the brightness/contrast filter:

[iconoclast]

Seems to look better, but I can't say that by seeing only that excerpt. And, of course, there is still that wrong lighting angle.

By the way, there is of course, also a psychological problem, because I know that it is a composing. So I'm searching for evidences. In such cases, it might possibly help to get statements from people who don't know that. If they see that there is something wrong, then you definetly know that you should change something.

And an additional annotation. In these times of HDR and oversharpening, photos often look somehow irritating and unnatural, at least in my opinion. Probably they are often intended to look that way. So possibly the problem with this image isn't as big as it seems to me.

[SolidSnake2003]

On 8/10/2023 at 1:40 AM, iconoclast said:

Seems to look better, but I can't say that by seeing only that excerpt. And, of course, there is still that wrong lighting angle.

By the way, there is of course, also a psychological problem, because I know that it is a composing. So I'm searching for evidences. In such cases, it might possibly help to get statements from people who don't know that. If they see that there is something wrong, then you definetly know that you should change something.

And an additional annotation. In these times of HDR and oversharpening, photos often look somehow irritating and unnatural, at least in my opinion. Probably they are often intended to look that way. So possibly the problem with this image isn't as big as it seems to me.

@TrentL Here is an updated look where I have added a shadow with lower opacity set to multiply mode: