Vincent Hector - Turntower

Turntower tutorial

This is a test project to embed a 3d model from a cad application directly into photoshop. Using adobe AI features to create the scene.

This workflow can be used in different ways and taken parts of it. Very easy to use an for some cases very useful.

In this case I used a 3d model created procedural with Itoo railcone, firefly for the scene, Photoshop/ substance viewer for 3d model into Photoshop and Photoshop generative fill to dress it up. No post production in this case for colours and levels.

This is just a use case and a bit of fun.

First set the scene and feel using adobe Firefly using a simple prompt.

Dutch landscape. Soft green, brown and grey tones. Some line with trees, river with lush green banks. canon camera, soft bokeh, ultra realistic, soft blue sky.

Firefly has a lot of options to adjust style, lighting and mood.

Now we want to control the scene using a block out out of photoshop. You can also use a cad image, illustrator file or just a photo.

At the composition tab in Firefly load the image. The strength can be set with the sliders.

In the reference style you can load an image what you like in atmosphere.

Load the result in and work with an other image to finetune it.

Result to photoshop for starting to place the 3d model.

For this i made quick a procedural model using Railclone. Model exported as a fbx, but a lot formats are compatible.

In photoshop you can place embed the file into photoshop. Double click on the layer and you will directed to the Substance viewer where you can change the light, angle, perspective and material.

Now that the 3d model is placed the composition feels a bit cramped.

With generative fill extend we can add some context to the top image.

The buildings feels like it is floating. Let’s use generative infill merge it.

Now put in some story and composition. Image needs a pier, some people and birds. lets try the generative fill again with prompts on area selections.

With blank Generative fill creating a person often goes wrong. So we can bring a photo of a person we like. With another ai tool object select we can remove the background. And with generative fill afterwards we can merge into the image.

Let’s see the result.

Page updated

Google Sites

Report abuse