For the retentive time , mental image generation was one of the few area whereGemini had an edge over ChatGPT . Google ’s Imagen model generated more naturalistic images and followed command prompt good , whereas OpenAI ’s DALL·E often produced images that take care more three-toed sloth - ish and cartoonish . But that ’s about to commute .
Now , both Google and OpenAI are rolling out native range of a function generators powered by their multimodal AI models . WhileGemini ’s versionis still tucked inside AI Studio — Google ’s genus Beta testing platform — OpenAI has release its aboriginal image source directly inside ChatGPT .
Here ’s the twist . Gemini ’s version is useable to everyone for free , while OpenAI delayed the rollout for free users due to excessive need . I test both . And allow ’s just say the tables may be turn .
Left: Original | Center: ChatGPT | Right: Gemini
Table of content
1. Changing the Style of an Image
One of the biggest perks of native image generation is that it ’s not just about creating images from clams — it ’s also great at editing or transform survive ones . After ChatGPT rolled out this feature , Xwas flooded with anime - style persona cheer byStudio Ghibli . So I make up one’s mind to try on the same . I upload a photo of a guy into both ChatGPT and Gemini , require them to commute it into anime Ghibli trend . Here ’s what I pay off :
Let ’s just say this clearly : Gemini scramble to change the style completely . It mostly stick around to the original image , often just bumping up the luminosity regardless of what you ask . In contrast , ChatGPT transform the image — anime dash , pixel art , lego characters , Simpsons , 3D pixar panache living , you name it . While it changes a few detail , particularly confront , the overall result is miles ahead of other poser .
Verdict : Only ChatGPT can amply translate an range into a different style . Gemini ’s aboriginal image generator currently light inadequate
Left: Original | Center: ChatGPT | Right: Gemini
2. Editing Small Changes on Top of Image
This is where Gemini glitter . I uploaded the same image to both models and asked them to add eyeglass . Here ’s what happened :
Both get down the line done — but in unlike style . ChatGPT tends to redraw the entire image , sometimes even altering the person ’s face . In demarcation , Gemini simply adds the glasses without touching anything else . That ’s because ChatGPT still sire a new image base on the original , whereas Gemini can make edits on top of the original . likewise , you may also slay objects , etc . in Gemini . mass are already using this lineament to move out watermarks and various wild stuff .
Verdict : Gemini win if you require clean edits without change the original figure .
Left: Original | Center: ChatGPT | Right: Gemini
3. Generating a Realistic-Looking Image From Scratch
This used to be a fallible spot for ChatGPT — but not any longer . I asked both manikin to generate a finale - up of an old human in his 70s , wearing a soft fleece cardigan over a white shirt . Here ’s what they make out up with :
Both nail the prompt and are middling much on par . ChatGPT ’s image is slightly more urbane , while Gemini ’s version feels more realistic , capturing subtle camera imperfectness and a natural looking . ChatGPT , by contrast , leans toward flawlessness — almosttooperfect at clock time . However , choose one over the other can be a personal preference .
Verdict : Both models can generate naturalistic images with repose .
Left: ChatGPT | Right: Gemini
4. Blending Two Images Into One
Both ChatGPT and Gemini now get you upload multiple reference photos to generate a fresh figure . In this exam , I uploaded a exposure of a humans and a disjoined simulacrum of another man wearing a green shirt . Finally , asked both models to generate a flick of the first military man wearing the unripe shirt . Here ’s what I incur :
ChatGPT deliver consistently skillful results . Gemini , however , sometimes skips the header or output a low - quality image — but those glitches normally go away with a retry or two . On the snotty-nosed side , Gemini nails the shirt color more accurately , while ChatGPT introduces a slight version .
Verdict : ChatGPT wins for consistency and overall character . But Gemini ’s not far behind
Left: Original | Center: ChatGPT | Right: Gemini
5. Generate an image with a Different Point of View
Both ChatGPT and Gemini can also give images from a different point of perspective . For this test , I uploaded a photo of a train ’s inside and postulate both model to recreate the aspect from the opposite side .
Both render enough results but struggled with object placement , especially in complex images with many elements . That said , if following such inside information is n’t authoritative and you ’re just take care for a sweet linear perspective of a subject — like a car or a building — both models cover it well .
Verdict : Both model did a decent job but messed with the arrangement of objects .
Left: Original | Center: ChatGPT | Right: Gemini
6. Generating a Birthday Card
These new models are also say to be much good at generating icon with text — a labor older model often struggled with . So , I decided to test that by generating a natal day board with specific textual matter .
Surprisingly , both models nailed it . They keep an eye on the command prompt on the dot — using cursive “ glad Birthday ” schoolbook and hem in it with floral designs , just as postulate . Which one await easily is really a matter of personal taste . We also tried generating menu , notice , infographics , and other text - sonorous images — and both model handle them quite well .
Verdict : Both Gemini and ChatGPT can now generate accurate , readable text within ikon and follow prompts closely .
Left: ChatGPT | Right: Gemini
ChatGPT vs Gemini Native Image Generator
Across all our tests , a few patterns stand out . ChatGPT consistently deport gamey - quality images — sharper details , well composition , and fewer weird artifacts . With Gemini , you often have to regenerate a few times before landing on something good . That said , Gemini is perceptibly faster . It can produce an image in around 10 mo , while ChatGPT can sometimes take a minute or more , even for childlike prompts .
When it amount to bring forth naturalistic visuals , handling text in images , or change perspective , both models perform similarly . But the real difference shows up during edits . ChatGPT is much honorable at transforming the overall style of an figure of speech , while Gemini glint when you want to contribute or remove specific objects without change anything else .
Overall : ChatGPT offer more uniform quality and a well all - around experience — if you do n’t mind the hold .
