Give Midjourney Clear Direction in 2026

THE FIRST PART OF YOUR PROMPT IS WHAT REALLY MATTERS

The Midjourney model continues to be refined and upgraded. Competition for text-to-image applications is fierce in today's market; there are dozens more vendors today than when Midjourney released it's first model, and its just getting ready to release version 8.

Learning to prompt the engine is key to getting good results. Right now, the best formula for getting good results is to give the engine very clear directions in the first couple of sentences of your prompt. The old days of long bloated prompts are no longer relevant. They confuse the engine and many of the adjective, adverbs and fluff words are not necessary to get good results.

Today, the best formula for getting good results is to formulate your prompt so the first sentence contains the subject, the action or what is happening in the image and role of the scene or where the subject is. Essentially, the "Who" "What" and "Where".

After Midjourney has this critical information, you can follow this with information such as:

lighting
atmosphere
symbolic elements
textures
camera feel
color palette
emotional tone
compositional depth
other important information you deem important

Using this very clear, concise format moves the most critical information for Midjourney to the front of the prompt and ensures that this information makes it into the engine. Each submission only gets a specified amount of time for the engine to interpret the prompt. Don't waste this precious time with unnecessary words up front.

When I built the prompts for the Mardi Gras whimsical stories, I asked CHATgpt to follow this structure.

That first sentence should include the subject, the action or what is happening in the image and role of the scene or where the subject is. Then the prompt can include supporting information with descriptive information about lighting, mood, clothing, anything else. But Midjorney will know "Who" "What" and "Where" from this first sentence.

I would like to build some whimsical scenes for Mardi Gras, as this celebration is coming soon, on February 17th this year. I don't need them too complicated, like our previous speed art we did back in January. I want to create them as composites in Photoshop. The Midjourney stories are for inspiration. Can you build me 5 test prompts in our new format.

Here are a couple of the prompts returned. You can clearly see that the "Who" "What" and "Where" are in the first sentence.

An elegant anthropomorphic black cat plays a trumpet in Jackson Square during a vibrant Mardi Gras celebration at night. Tail curled dramatically, emerald waistcoat and purple tailcoat jacket, gold embroidery catching warm gas lamp light, swirling musical notes subtly glowing in the air, soft haze from street mist, cobblestone textures beneath, lively but dreamy atmosphere, cinematic lighting with high contrast highlights and deep shadows, rich saturated color palette.

A mysterious Mardi Gras queen adjusts her ornate mask while standing beneath hanging beads and lanterns in a decorated French Quarter courtyard. Lavish layered gown in deep violet with gold filigree details, cascading bead necklaces, elaborate feathered headdress, marble fountain softly visible behind her, glowing lantern light mixed with cool moonlight, drifting glitter particles in the air, subtle wind movement in fabric, elegant yet whimsical tone, painterly textures, rich jewel-tone palette.

I would then break down these stories into component and render the background, the character and some elements all separate so that I can build a composite in Photoshop. Here is an example of a completed artwork from this process.

Categories: : Midjourney