Image credit: @jacksonstockphotography

Midjourney V5 Prompts & PromptBase

Today Midjourney publicly released version 5 of its text-to-image model.

Midjourney has proven itself to be the model of choice for image-focused prompt engineers. Since launching version 4, it quickly became the most popular model on PromptBase, overtaking DALL·E and Stable Diffusion. Currently, 60% of all sales on PromptBase are Midjourney prompts.

The range and quality of images Midjourney can produce when prompted well is remarkable. We are very excited to see what creators on PromptBase will be able to achieve with this new model.

This blog post will highlight some of the changes we've made across Midjourney prompts on PromptBase, and how to approach this latest version update.

Version tags

On all Midjourney prompts we now display the version of Midjourney that was used for the prompt.

We strongly believe that it doesn't matter what model you're using, as ultimately the final image quality and prompt quality is what matters the most. However, we also appreciate that users may want to know the model before purchasing, so this serves as an easy way to determine this.

Updating prompts

Midjourney states that currently version 5 is much more opinionated than v3 and v4, and that shorter prompts that once worked may not work as well.

Therefore, it is crucial that if you are updating your existing prompts from v4 to v5, you ensure that it still produces images that reflect the idea behind your prompt. This may involve significantly editing your prompt.

If you do decide to update prompt versions, make sure to submit new example images generated by your version 5 prompt.

It is not necessary to update your prompt - it will not affect its listing in the marketplace. If your prompt works well on v4 we strongly recommend keeping it on that version.

Midjourney V5 alpha

Midjourney V5 is currently in alpha, and Midjourney state that significant changes could be made to the model in the future.

Creators using V5 should ensure any prompts listed that are created using V5 remain consistent over time. If Midjourney sunset this current version of V5 to a model tag like --v 5-alpha, then we will update all your existing prompts to this model version.

Pro mode & new pricing tiers

Midjourney state that version 5 is currently in 'pro' mode by default, stating "It’s MUCH more ‘unopinionated’ than v3 and v4, and is tuned to provide a wide diversity of outputs and to be very responsive to your inputs".

Some of the most talented prompt engineers are listing prompts on PromptBase, so this statement excites us greatly, as users can now push the limits of what is possible with Midjourney with no guard rails.

As such, we have decided to add additional prompt pricing tiers up to $29.99 across all prompts. This also coincides well with new GPT-4 prompts being able to provide more value in advanced applications.

Midjourney state "We hope to have a ‘friendly’ default styling for v5 before we switch it to default", which indicates a --raw or a --pro tag coming soon. When/if this switch happens we will append this tag to all existing v5 prompts submitted to PromptBase during this alpha period.

Conclusions

It has been an exciting week for prompts! We have had not one, but two state-of-the-art models released to the public.

The most interesting thing with Midjourney V5 is that, even though it is a much more advanced model, the requirement for prompting well actually increases instead of decreases.

As prompts and models get applied to more and more domains (music, video etc.) with even more advanced models, we see this as a signal that the emergent field of prompt engineering is here to stay around for a while.

Those users who can express themselves clearly, and draw upon pockets of specialised knowledge from various domains will excel at getting the most out of prompting these advanced models.

At PromptBase, we are excited to enable these multi-disciplinary super creators!