Inserting Content Into Images

Most typical intervention on our renderings is adding content such as plants and people. Here's how to do it like a pro!

Whenever you present the architectural space through 3D rendering, you will most probably need to fine-tune the rendering and  add a large amount of content to it. This is a process called “image compositing” which means that your final image will be sum of your rendering, and bunch of other images you used to bring this rendering to life.

In order to perform these tasks, there are a lot of (possibly) new concepts that you should learn about Photoshop:

But don’t worry, we will not focus on showing you the “features” of Photoshop.
Instead we will be focusing on the actual workflow – this tutorial will show you how to get things done. There are still a lot of Photoshop functions and features that you have to get used to, but it’s easier to learn when you actually keep applying these things in you real work.

So let’s get started.

Download the tutorial files

I  have prepared several images for you to work with in this tutorial, and you should download them from here:

Download The Tutorial Files (14.3Mb)

In this tutorial we will start from the image on the left and add content to create something similar to the image on the right:

base-imagefinal-image

Getting The Content Into Your Images

Getting the content such as people, cars, plants is usually straightforward copy + paste, as long as the content image is nicely cut out with transparent background (which in Photoshop shows as a checkered pattern). An example of a transparent image:

transp

If you don’t have already cut-out images, you will have to spend a lot of time cutting them yourself (which will be a subject of another tutorials). Luckily enough, there are a lot of such images to be found online (check 2A post > Resources), and you will also be provided with large library of these images (again, check the same post).

Inserting New Content

So let’s get started. Here’s how to get this elderly couple into your base image (which can either be a photo or a rendering – most often you will be working with renderings):

Free Transform = Resizing/Transforming Content

Obviously, the people we inserted are way too large for this composition. So let’s resize them and put in the suitable place:

Perspective deformations

Very often you will need to transform the new layers in such way that they fit the perspective of your main image. To show you how, we’ll try to add “roof” to one of the boxes, like this:

Fixing Visibility Issues

Most often the content we want to insert will be only partially visible, and partially “behind” something we already have in our scene.

Easiest way to go about it is to just select a part of the new layer that should be “hidden behind” and delete it, as in this example of putting a tree inside of one of the boxes. Here’s how to do this in a quick-and-dirty way:

That was quick and easy, but the reason I said “quick and dirty” is that this process permanently deleted a part of our tree image. If we made a mistake or just want to move/resize tree a little differently, we may realize we get to need the part we just deleted. So, in order to work like a pro, you should embrace the use of layer masks.

Use Layer Masks To Work Like A Pro

Layer masks are one of the most important and often-used features of Photoshop, because they allow us to hide parts of layer without deleting them.

A layer mask is a greyscale image “attached to a layer” that defines the visibility of parts of that layer. White parts of mask = layer is fully visible at these positions. Black parts = fully hidden, with all the shades of gray in between.

It works very similar to “transparency texture” in Rhino: Each pixel on a layer mask correspond to a pixel on its main layer. Parts of the masks that are white will keep these parts of the layers fully visible. Parts of where the mask is black, the main layer will be hidden. And all the greys in between will affect the visibility/transparency accordingly.

Here’s how to select things and add a layer mask:

Please note that here I also introduced the concept of History. By default, Photoshop has only one undo – but if you look at history panel, you can see everything you’ve done on your document, and revert these changes as you see fit. Keep that in mind because you will be using History a lot!

Moving layer independently from the mask.

This is another great concept that masks allow us. By default, mask is attached to its layer meaning they will move together. However, if we un-chain them, then we will be able to move both of them independently.

Why is this important? Well very often we want mask to stay in place and move the layer only – in this case, our mask defines the “edge of the cube” and these should not move when we try to move/resize the tree around to maybe better fit our intended design.
Here’s how:

These are two most basic and most important workflows related to Layer Masks.
My strong recommendation is that you should start practicing masks as soon as possible and learn to use them quickly – it may seem complicated at first, but it’s something that will greatly improve your productivity, almost like no other Photoshop feature.

 

Organizing Photoshop Files: Name Your Layers

Talking about productivity, the best practice that will make your work easier is to give each layer a name that is clear to understand. Very often your Photoshop files will grow to have dozens and maybe even hundreds of layers, and it’s going to be hard to find the layer you want to work on, unless you keep your layer structure neatly named.
So let’s do it now (although this file will remain fairly simple):

Blending The New Content With The Base Image

The most common mistake the beginners make when doing image compositing is that they fail to make the new content look like part of the base image. This means that new content will have different color hues, or even worse – very different lighting than the base image, and as a result the entire result will look very artificial and clumsy.
For example:

blend Matching lighting conditions: an example of well-placed and terribly placed content.

It is obvious that the women in green t-shirt looks terribly out of place in this document because -primarily- the sunlight on her is coming from another direction (we should flip (mirror)  that layer to get the good sun orientation). Another reason why it doesn’t look good because the color scheme is very different from the base image.

Here’s how to fix the two:

Painting The Shadows Manually

Image compositing usually involves a lot of manual painting, because the new content will rarely look perfectly lit to fit our scene. Luckily enough, this is not difficult to do. Here’s how:

Basically, we added a new layer above our tree, and painted the shadows that would be expected to appear if this tree was really inside of the box.
Important: See how the layer mask for this new shadow layer is created, in order to limit our painting only to the tree (and avoid painting the background behind).

Pro tip: You can also paint light, not only the shadow!

Matching The Colors

As I said above, another typical thing we need to do is to change the color scheme of our new content slightly to make it blend better with the background. We can simply use Hue/Saturation interface to shift the colors around a bit, and (usually) decrease the saturation of parts of our new layers. Here’s how:

Pro tip 1: Often our content will blend better if we make it a little bit transparent (ie. reducing opacity to around 85%).

Pro tip 2: Don’t forget that in Hue/Saturation interface you can use channels to affect only some of the colors (ie. desaturate reds while keeping the blues saturated). This was explained near the end of this video.

 

Putting It All Together

Finally, let’s apply all of this again at once, in order to add a “bush” on the new roof we created, and adjust its lighting and color conditions to make it blend better:

So here’s the final image in all its -rather confusing- glory:

final-image

 

 

 

 

Any questions?

Leave a Reply

You must be logged in to post a comment.