6 ideas, tips, and prompts for creating photos with Grok AI

When X, formerly Twitter , launched Grok 2 , the AI turned shortly identified for its snarky humorousness and its capability to generate photos with fewer restrictions than different platforms. However, Grok 2 additionally tends to generate extra sensible photos than competitors like DALL-E . That, combined with the $8 month-to-month entry to X Premium that is still decrease than Google and ChatGPT’s $20 a month AI subscriptions, has earned the software program a popularity earlier than even leaving beta.

Nevertheless, profiting from the platform’s capabilities requires a little bit of know-how to get the very best outcomes. Like most AI platforms, the way you phrase the immediate issues. However, with Grok’s looser restrictions, customers ought to perceive not simply the best way to use the AI, however the best way to use the AI in probably the most moral method potential.

Earlier than you begin typing in that immediate field, these 6 tips and ideas will assist you to benefit from the brand new X AI.

1 Be aware of AI ethics

Grok can do issues it in all probability should not…

X / Grok

X describes Grok as an AI with “a touch of riot,” which loosely interprets right into a chatbot that has fewer guidelines, notably regarding picture technology. Not like DALL-E or Google Gemini , Grok can generate photos of recognizable individuals, replicate the type of a particular artist and even a particular piece of art work, and reproduce different mental property together with logos and licensed characters.

However, simply because Grok can do these issues does not imply you ought to. Even when wielding Grok and a keyboard as a substitute of a paintbrush, you danger lawsuits infringing on mental property when replicating an artist or together with issues like logos or licensed characters. Creating photos of recognizable individuals might danger violating Right of Publicity in relevant states, laws that forestalls somebody’s likeness from being commercially exploited.

Some can also select to keep away from AI in any respect — or be selective on which platforms to make use of — over concern with how the coaching knowledge was gathered. Artists have beforehand taken AI corporations to court docket over utilizing their work to coach knowledge. xAI says Grok 1’s initial training data got here from “a wide range of textual content knowledge from publicly obtainable sources from the web,” nevertheless it would not specify the place its picture coaching knowledge got here from. Grok 2 makes use of the FLUX 1 mannequin by Black Forest Labs, which is equally unclear on the place the coaching knowledge originated from. The inclusion of “signatures” in generated art work suggests copyrighted work could also be a part of the information.

The ethics of utilizing any AI platform apply to Grok as nicely, together with correctly labeling the picture as AI generated.

2 Embrace Grok 2’s strengths and perceive its weaknesses

Keep away from asking for photos with close-ups of fingers or graphics with textual content

The important thing to getting the very best outcomes from any AI is to know the platform’s strengths and weaknesses, and Grok isn’t any exception. Grok’s strengths lie in its extra sensible picture technology, its capability to deal with duties like producing photos that appear like celebrities, and its deal with humor. When evaluating Grok to DALL-E, I’ve discovered Grok produced photos that had been extra sensible (although nonetheless had a number of tell-tale AI indicators).

However, Grok 2 is a younger program nonetheless in beta, and it has some limitations, regardless of its popularity for having looser content material pointers. Whereas Grok itself will let you know that it is multimodal, it is not, or a minimum of not but. There is not an choice to connect a picture to ask Grok so as to add one thing to it or generate one thing primarily based on that file. It is a text-only enter at present.

Whereas Grok typically delivered photos that regarded extra superior than DALL-E, the AI nonetheless faltered at among the duties that appeared to be universally tough for AI. Palms and fingers can typically look wonky. Textual content is commonly gibberish and even made up of non-letters until you explicitly spell out what to say.

There are additionally some directions that Grok would not appear to know but. For instance, I could not get Grok to generate a picture in a particular side ratio, regardless of explicitly asking for 16:9, I needed to go in and crop to that format afterward as a substitute. When asking for photos of individuals, Grok tends to make all of them appear like runway fashions until requested in any other case.

3 Write particular prompts, somewhat than generic

The extra detailed, the higher

X / Grok

Like most AI platforms, Grok delivers the very best outcomes when given a particular, well-worded immediate somewhat than a generic one. As you kind within the textual content field, transcend simply what individuals or objects ought to be within the picture. Think about elements like:

• Fashion: Grok can create photos that look sensible, or you may ask for a cartoon or perhaps a portray. Think about in order for you vivid, daring colours, or extra muted tones.
• Temper: Specify what kind of tone you’re searching for, whether or not you desire a humorous picture, a darkish depiction, or maybe one thing skilled.
• Background or location: Inform Grok the place this image ought to be positioned with a view to get a background that matches your imaginative and prescient. Be particular with settings, like a espresso store or a busy avenue.

4 Ask for tweaks till you get the very best end result

Grok takes your prompts in context

Grok would not neglect your earlier immediate, which suggests when you don’t get a very good end result, you may inform the AI extra particulars. You do not have to re-type the immediate. As an alternative, say, “Regenerate that however with _____.” If you do not have extra textual content to make clear the end result with, you may hit that round arrow key beneath the unique technology to regenerate from the identical immediate.

Some phrases additionally work higher than others, so chances are you’ll need to experiment with rewording your immediate, somewhat than including extra particulars. After I requested for a “various” group for instance, it generated a picture of individuals all with the identical darkish pores and skin tone. As soon as I requested for “a wide range of pores and skin tones,” then the AI created a gaggle of ethnically various individuals.

On the flip facet, if you wish to create a wholly new picture, typically Grok will nonetheless pull particulars from the earlier immediate, just like the background or type that you just requested earlier. If this occurs, merely begin a brand new chat thread so Grok “forgets” your earlier requests.

5 Have enjoyable with Grok’s snarky humorousness

Grok was made to be a much less critical AI

X / Grok

X’s new AI is thought for each its free content material pointers and its humorousness. Embrace the humorous facet of Grok as you experiment. Grok isn’t a digicam, so create off-kilter pairings that you just wouldn’t see in actual life to make an announcement, like a sandwich with a facet of on line casino “chips” or a sensible rendering of the expression “raining cats and canines.”

6 Test your privateness settings

X can use your knowledge to tweak Grok

According to X, Grok was pre-trained on a wide range of knowledge pulled from the web, not public X posts or Tweets. Nevertheless, a data-sharing setting is now routinely turned on by default for X customers, until they opt-out. This setting says that “we might make the most of your X posts in addition to your person interactions, inputs, and outcomes with Grok for coaching and fine-tuning functions.”

If you don’t need Grok to be skilled in your posts you may decide out by tapping the three-dot or extra menu, choosing Settings & Privateness, clicking Privateness and Security, then choosing Grok. Then, uncheck the field to opt-out.

Grok additionally makes use of public Tweets to share real-time knowledge with customers. If you don’t need your posts probably popping up in Grok, you will have to make your account personal by going to Settings and Privateness > Privateness & Security > Viewers and Tagging > Defend your posts.