ChatGPT vs. MidJourney: Which AI Art Tool Creates Better Images?
Table of Contents
- Comparing Realistic People Images
- Adding Convincing Text to Images
- Built-in Image Editing Capabilities
- Creating YouTube Video Thumbnails
- Key Differences in Pricing and Ease of Use
Comparing Realistic People Images with MidJourney and ChatGPT
Until recently, MidJourney was regarded as the number one AI art generator while ChatGPT was only used as an AI writer. But with ChatGPT’s latest DALL-E 3 update, you can now make AI art directly in ChatGPT, putting the two platforms in direct competition.
Our first test focused on one of the biggest challenges for AI art platforms - generating realistic-looking people that don't end up looking strange or creepy. Over time, MidJourney has evolved from creating very unrealistic people to now generating images that could pass for real photos used in ads or websites.
We first tried a simple prompt in MidJourney asking for a realistic photograph of a person. Most images looked decent at a quick glance but seemed digitally altered upon closer inspection. However, one image on the bottom right looked like an actual photograph. We then tried a more complex prompt with additional details about photographic style, lighting, etc. This resulted in an extremely convincing image that would likely fool photography experts.
We then tried the same test in ChatGPT. With a simple prompt, the AI-generated person looked more like a drawing than a photograph. But with a more detailed prompt, ChatGPT created an exceptionally realistic image. The pores, lines, and other finer details make the image seem real on close inspection.
For simple prompts, MidJourney delivers better realism. But with more complex prompts, MidJourney and ChatGPT deliver similarly convincing results, demonstrating ChatGPT's strong capabilities given sufficient detail.
MidJourney Performance
MidJourney has greatly improved at generating realistic people over time. Simple prompts still sometimes result in unrealistic or 'off' images. But with more detailed prompts that include descriptors of photographic style, lighting, etc., MidJourney can produce highly photorealistic images. The evolution of MidJourney's capabilities with detailed prompts allows it to compete with the realism of other leading AI art platforms like ChatGPT.
ChatGPT Performance
With simple prompts, ChatGPT's image generation capabilities result in drawings or paintings moreso than photorealistic images. But the latest DALL-E 3 update integrated into ChatGPT allows for exceptionally realistic images given sufficiently detailed prompts. ChatGPT is fully capable of generating convincing images comparable to dedicated AI art platforms. With the proper prompts, ChatGPT delivers professional image quality and realism.
Comparison
For simple prompts, MidJourney generally produces more realistic images. But both platforms are capable of generating highly convincing photorealistic people given detailed prompts that describe scene composition, lighting, photographic style, etc. MidJourney likely still has a slight edge for ease of generating professional images. But ChatGPT delivers exceptional quality with sufficient prompting and is far easier to use overall.
Adding Convincing Text to Images with MidJourney and ChatGPT
AI-generated images often struggle with accurately rendering text, which is essential for applications like graphic design, social media posts, etc. We tested how well MidJourney and ChatGPT handle adding custom text to images.
MidJourney has notorious difficulty properly generating text, often leaving out letters, using non-standard alphabets, or completely misspelling words. We prompted MidJourney to create images like stop signs, but text accuracy remained an issue.
ChatGPT performed surprisingly well at incorporating custom text into images. Some letters were missing or inaccurate on complex images, but ChatGPT overall handled text in a much more usable way than MidJourney currently does.
For most practical purposes, ChatGPT's text rendering vastly outperforms MidJourney's text attempts. Those needing to incorporate text into AI art should use ChatGPT or edit MidJourney images in a design platform like Canva.
MidJourney Struggles with Text
When prompted to add text, MidJourney regularly leaves out letters, uses odd versions of alphabets, or completely misspells words. The surroundings of images look good, but accurately rendering readable text remains extremely difficult. Using MidJourney-generated images with text for professional or commercial purposes would likely require post-processing in image editing software to fix text inaccuracies.
ChatGPT Handling Text Well
While some minor text inaccuracies persisted, especially on complex images, ChatGPT overall produced remarkably usable text within images without needing further editing or post-processing. Needing to add custom text no longer precludes choosing an AI art platform. ChatGPT appears more than capable of handling text at a quality usable for professional creative projects.
Built-in Image Editing Capabilities
A key consideration in choosing an AI art platform is built-in editing tools allowing you to iteratively improve images without needing to start over from scratch each time. We explored MidJourney’s and ChatGPT’s respective editing capabilities.
MidJourney provides several ways to edit images like varying regions, expanding scope, and upscaling quality. These tools enable easily iterating on an initial image to get closer to the desired result, saving huge amounts of time and effort.
ChatGPT currently has minimal editing functionality. Small batches of variations can be requested, but no direct manipulation of a chosen image is possible. Any changes basically require generating new images from scratch rather than editing.
MidJourney clearly provides superior direct image editing tools at this time. Quick iteration on ideas is far easier than with ChatGPT’s limited post-generation options.
MidJourney Editing Options
MidJourney allows editing chosen images in multiple ways, like varying specific regions, expanding outward, moving viewpoints, and upscaling quality. These direct editing tools enable quickly iterating on initial AI-generated images to refine ideas. Built-in functionality for cropping, transforming, and manipulating areas of interest provide creative flexibility and save huge amounts of time over generating completely new images.
Limited ChatGPT Editing
ChatGPT currently lacks any real editing capability for AI-generated images. Small batches of variations can be requested, but no directly editing a chosen image is possible. Effectively, new ChatGPT images must be created from scratch each time rather than refining ideas through intentional manipulation as with MidJourney’s more advanced toolset.
Creating YouTube Video Thumbnails
As a final test of capabilities, we challenged MidJourney and ChatGPT to each create a YouTube video thumbnail with AI-generated elements showing two robots facing off, representing the platforms themselves.
ChatGPT quickly produced nice images that could work as thumbnails with some added text. MidJourney's initial attempts didn't match the desired concept. Step-by-step building was required, first generating a background, then robot characters in suitable styles, and finally compiling assets in Canva.
While more effort overall, MidJourney ultimately enabled creating a polished, professional thumbnail. But ChatGPT delivered strong initial thumbnail options easier and faster. Both platforms have merits for creative projects, with MidJourney potentially enabling more complex executions.
MidJourney Step-by-Step Process
Creating a multi-element YouTube thumbnail was far more involved with MidJourney compared to ChatGPT. A background graphic and two robot images had to be generated separately based on iteration and prompts. Compiling all elements into a cohesive thumbnail ultimately required exporting assets and constructing the full image in Canva. More complex executions demand planning, but polished results are achievable.
ChatGPT Solid Initial Attempts
With a simple prompt, ChatGPT produced several thumbnail options showing robot characters facing off that could work nicely with only minor added text or refinement. Less iteration on discrete elements was needed upfront. While less flexible for highly complex images, ChatGPT enabled respectable results faster and with less effort.
Key Differences in Pricing and Ease of Use
MidJourney and ChatGPT take different approaches when it comes to pricing models and overall usability.
ChatGPT requires a paid ChatGPT Plus subscription which unlocks DALL-E art capabilities along with other features like web browsing. The combined creative tools make the $20 monthly fee reasonable for casual users.
MidJourney pricing starts at $10 per month. The sole focus on art generation means producing more images faster. But understanding MidJourney's immense number of parameters and options poses a learning curve.
Ease of use favors ChatGPT. The conversational interface feels beginner-friendly even with limited art experience. MidJourney's complexity caters more to power users willing to invest time mastering prompts.
FAQ
Q: Which AI art tool is better for beginners?
A: Currently, ChatGPT's DALL-E integration is more beginner-friendly and requires less precision to generate quality images.
Q: What editing options exist in MidJourney versus ChatGPT?
A: MidJourney has more built-in editing tools like varying regions, zooming out, and panning. ChatGPT has minimal editing capabilities.
Q: How do the pricing models differ between platforms?
A: MidJourney starts at $10/month just for art generation. ChatGPT is $20/month but includes art plus other features.
Q: Which created more realistic looking people?
A: With simple prompts, MidJourney won. But with more advanced prompts, ChatGPT created very convincing images on par with MidJourney.
Q: Did either platform properly handle text on images?
A: ChatGPT showed better text handling. MidJourney struggled with missing or incorrect letters.
Q: Were thumbnails successfully created in both tools?
A: Yes, both ChatGPT and MidJourney were able to create YouTube thumbnails after a few attempts.
Q: Is one tool clearly better overall?
A: It's hard to declare one platform as clearly superior. Each has strengths and downsides regarding capabilities, ease-of-use, and pricing.
Q: What are the benefits of MidJourney?
A: MidJourney offers more control for advanced users, built-in editing features, and lower cost just for art generation.
Q: What are the benefits of ChatGPT's DALL-E?
A: DALL-E is simpler for beginners to use, handles text better in images, and is integrated with ChatGPT's other features.
Q: Which platform will be better long-term?
A: Only time will tell! Both MidJourney and ChatGPT are rapidly evolving their art generations tools. The future landscape remains dynamic.