I started writing again after a long hiatus and quickly realised that creating compelling blog covers is one of those tasks that can easily become a bottleneck in your workflow. You want something that looks professional, matches your vibes, and captures the essence of your post, but you don't want to spend hours in design tools or may not have the skills to do it yourself. After experimenting with various approaches, I've settled on a system that gives me beautiful, consistent cover images in minutes: **parameterized AI prompts**. Let me walk you through exactly how I do it. ## The Prompt I use [ChatGPT's Image Generation](https://openai.com/index/introducing-4o-image-generation/) feature with a reusable parameterised prompt that goes like this: ```bash 4K (high-resolution) 16:9 abstract blog header — gradient blend of {PRIMARY_COLOR}, {SECONDARY_COLOR}, {ACCENT_COLOR} Texture: {TEXTURE_STYLE} with a subtle {MATERIAL_FINISH} sheen Gaussian-blur strength {BLUR_INTENSITY} (0 = none) Film-grain strength {GRAIN_INTENSITY} (0 = none) Geometric overlay: {GEOMETRY_OVERLAY} Lighting mood: {LIGHTING_MOOD} Modern, luxurious, minimal aesthetic, no text ``` The magic is in the parameters. Each one controls a specific aspect of the generated image, allowing me to create variations that feel cohesive but unique. Here's how each parameter works | **Parameter** | **Type / Range** | **Allowed / Typical Values** | | ------------------ | ---------------- | ----------------------------------------------------------------------------------------------- | | `PRIMARY_COLOR` | Hex or CSS color | Any hex code (`#RRGGBB`) or valid CSS color name (e.g. `#0099ff`, `midnightblue`) | | `SECONDARY_COLOR` | Hex or CSS color | Same as above | | `ACCENT_COLOR` | Hex or CSS color | Same as above | | `TEXTURE_STYLE` | Enum | `golden_sand`, `brush_strokes`, `mystic_smoke`, `marbled_ink`, `fluid_clouds`, `geometric_wave` | | `MATERIAL_FINISH` | Enum | `silk`, `velvet`, `matte_paper`, `frosted_glass`, `metallic_foil` | | `BLUR_INTENSITY` | Float 0 – 1 | `0` = none • `0.25` = subtle • `0.5` = soft • `1` = heavy | | `GRAIN_INTENSITY` | Float 0 – 1 | `0` = none • `0.1` = light film grain • `0.5` = noticeable • `1` = very coarse | | `GEOMETRY_OVERLAY` | Enum | `none`, `four_square_tiles`, `subtle_circle_cluster`, `diagonal_ribbon` | | `LIGHTING_MOOD` | Enum | `soft ambient`, `radiant`, `moody`, `high key` | Once you choose the parameters, simply drop the resulting prompt into ChatGPT or other AI image generation tools and you will get some interesting results. ## Let's talk results To see how versatile this system is, here are several examples of prompts I've generated using different parameter combinations, along with the beautiful images they produced: #### Mystic Purple with Velvet Texture #### Deep Ocean with Frosted Glass #### Solar Flare with Matte Paper #### Arctic Aurora with Silk Sheen #### Cyber Neon with Metallic Foil #### Desert Warmth with Golden Sand #### Pastel Dreams with Silk Texture #### Liquid Silver with Frosted Glass #### Cosmic Starlight with Metallic Foil #### Volcanic Fire with Velvet Texture #### Abyssal Depths with Frosted Glass #### Polar Aurora with Silk Sheen #### Opalescent Pearl with Matte Paper #### Peach Blush with Silk Clouds As you can see, the results are quite impressive considering the time it took. I decided to take this approach one step further and built [**PromptSauce**](https://www.promptsauce.com/) - an online directory to search for and share your design assets and the prompts used to generate them.

Blog

Recent Posts

Blog covers powered by GPT-4o

PG 101 - Building Postgres Extensions

Hello World