You will never have issue with finding right images again.

Greg Vonf @ Business Automated
7 min readFeb 24, 2023

--

3 strategies I use to accelerate content creation, by creating AI art assets at scale using OpenAI Dall-e, Airtable and Make.

Side by side comparizon of photo by Jadson Thomas from Pexels and AI generated image variation by OpenAI Dall-e

OpenAI Dall-e is an artificial intelligence (AI) model developed by OpenAI which is capable of generating creative images based on text descriptions. This model has the potential to revolutionize content creation by allowing anyone to quickly generate high-quality images for their projects. With OpenAI Dall-e, users can input a text description of what they would like to create, and the AI model will generate a corresponding image. This can be a great time-saver for anyone who needs to create images for their content quickly and easily. Additionally, the model is capable of creating unique and creative images that may not have been possible to create manually. This opens up a world of possibilities for content creators who want to create unique and eye-catching visuals for their projects.

In this article, I will discuss three strategies I use to accelerate content creation by creating AI art assets at scale using OpenAI Dall-e, Airtable and Make. With the help of these tools, I am able to quickly generate high-quality images for my projects. Additionally, I have the ability to create unique and creative visuals that may not have been possible with manual methods. Finally, I will provide an example of the results of OpenAI Dall-e, so you can see the potential of this powerful AI tool.

Tools needed

The 3 tools I am using here are Airtable, Make.com (formerly Integromat) and OpenAI API. These allow content creators to quickly create and process large amounts of prompts, automate workflows and generate creative images.

Airtable

Airtable is a powerful cloud-based spreadsheet software that allows users to create and manage their data in an organized manner. It’s a great tool for content creators who need to create and manage large amounts of data quickly and easily. With Airtable, you can create custom databases with fields, share them with your team, and sort and filter data with ease. This makes it an ideal tool for content creators who need to quickly create and manage large amounts of data.

Make.com (formerly Integromat)

Make.com (formerly Integromat) is a powerful automation tool that allows users to quickly create automated workflows. This can be a great tool for content creators who need to quickly generate large amounts of data. With Make.com , users can connect different services, such as Airtable and OpenAI, and create automated workflows that will generate images based on text descriptions. This can be a great time-saver for content creators who need to quickly generate images for their projects.

OpenAI API

OpenAI is an artificial intelligence (AI) model developed by OpenAI which is capable of generating creative images based on text descriptions. This model has the potential to revolutionize content creation by allowing anyone to quickly generate high-quality images for their projects. With OpenAI, users can input a text description of what they would like to create, and the AI model will generate a corresponding image. This can be a great time-saver for anyone who needs to create images for their content quickly and easily. Additionally, the model is capable of creating unique and creative images that may not have been possible to create manually. This opens up a world of possibilities for content creators who want to create unique and eye-catching visuals for their projects.

Build faster! Get the templates and Make blueprint here:

https://businessautomated.gumroad.com/l/generate-images-with-openai-dalle

Strategy 1: Images generated from prompts

The first strategy I use is to generate images from prompts. This involves creating a database of prompts in Airtable, which can then be used to generate images with OpenAI. This is a great way to quickly generate high number of images for variety of prompts.

Layout of Table to view multiple prompt generated images

The base is linked with Make.com (formerly Integromat) and when Status is set to Generate, the scheduled make automation places call to Open AI API to create and retrieve images. Important — you need to also set number of images you want to generate in the column number.

Scenario number 1 places call to Open AI API with a direct request and returns images

You can see details of the API calls and how to set up the custom HTTP call in the video linked at the end of this chapter.

Make sure that the prompt does not contain forbidden topics as per OpenAI policy as well as strange formatting artefacts which might case error in the response.

Watch full tutorial here:

Strategy 2: Creating variations of an exiting image

The second strategy I use is to generate variations of existing images. With OpenAI, users can input an image they already have, and the AI model will generate variations of this image. This is a great way to quickly generate a large number of variations of an existing image, which can be used to create more interesting visuals for content This opens up a world of possibilities especially if you have a good reference image.

In order to generate variations of an existing image using OpenAI, users need to provide the model only with an image. The requirements are:

  • PNG format
  • Less than 4MB
  • Square aspect ration
Example of table in our base to create image variations

Similarly to step above we need set the number of images to create and set the Status field to “Generated”. Based on that the Search module in Make.com will look for all rows with that status and send data (original image) to OpenAI for processing. Once results are return, these are being uploaded to Airtable

Make offers a no-code way of connecting to Open AI API to create image variations

For a full walk-trough and API call details you can watch this tutorial:

Strategy 3: Editing existing images

The third strategy I use is to edit existing images. With OpenAI, users can input an image they already have, a mask indicating area to change, as well as a prompt describing the final scene.

Based on these 3 inputs the AI model will generate edited versions of the original image. There are a few important requirements about the input image and the mask:

  • Both image and mask needs to be under 4 MB and in the PNG format
  • The images need to be square and have the same dimensions

The mask is created by adding opacity (transparency) to original image. You can for example:

  • Keep the central object in place and remove background. I use https://photokit.com/ and cut out tool for that.
  • You can remove the object in the scene and keep the background
  • You can also selectively add partial opacity e.g. 50% which you want to partially modify. The higher opacity , the more of original image will be retained by AI.
Edit image table structure

As the next step you need to select number of images you want to generate, it must be between 1 to 10. Change status to “Generate” and trigger Make scenario.

Make scenario contains 3 custom calls in the middle. First we are downloading the base image and mask and uploading this to Open AI.

Make scenario downloading the image and mask for OpenAI to create edits.

Make scenario downloading the image and mask for OpenAI to create edits.

You can see details of the API calls and how to set up the custom HTTP call in the video linked at the end of this segment.

Once Images are uploaded to OpenAI, Dall-e model processes them and returns files which are uploaded back to Airtable, while the Status is being changed from “Generate” to “Done”.

This model API is what is used for Dall-e outpainting feature, which allows to expand background of existing images and create bigger images of any shape and width. To build feature like that, you would need to execute multiple API calls with partial image from the edge of original picture and 0 opacity in places where the completion is supposed to happen.

I hope this scenario was useful for you. Let me know in comments how do you see these scenarios being used in your business and what AI features could help you!

Watch the full video tutorial of this scenarios here:

Build faster! Get the templates and Make blueprint here:

Business Automated is an independent automation consultancy. If you would like to request custom automation for your business, visit us at https://www.business-automated.com

If you like our tutorials — buy us a coffee☕: https://www.buymeacoffee.com/business

Follow us on Twitter🐦: https://twitter.com/BAutomated

Watch more on Youtube ️📺: https://www.youtube.com/c/BusinessAutomatedTutorials

--

--

Greg Vonf @ Business Automated

Greg is the founder of Business Automated, an agency helping small businesses streamline and simplify their processes. For more visit www.business-automated.com