in

Revitalizing Outdated Computer Graphics Using AI Image Generation – Jay Alammar – Visualizing the World of Machine Learning Concept by Concept



**Can AI Image Generation Tools Enhance Old Video Game Graphics?**

**Remaking Old Computer Graphics with AI Image Generation**

One of the fascinating capabilities of AI image generation tools is the ability to recreate and enhance old video game graphics. In this article, I will explore the process and my experience in using commercial AI generation tools to retell a story with higher fidelity graphics. Specifically, I will focus on the intro cinematic of the video game Nemesis 2 on the MSX.

**Meet Dr. Venom – The Original Image**

Dr. Venom, the villain in Nemesis 2, appears in the intro cinematic of the game. The image of Dr. Venom in this cinematic is a dramatic reveal. Now, let’s update these graphics using visual generative AI tools and compare the results.

**Comparing Original and AI-Generated Images**

To compare the original images with the ones generated by AI tools, let’s take a side-by-side look at the panels:

*Panel 1: Fighter Jets over a Red Planet*

The final image was generated by Stable Diffusion using Dream Studio. However, getting to this image involved generating over 30 images and tweaking prompts. The prompt used for Dall-E and Stable Diffusion was “fighter jets flying over a red planet in space with stars in the black sky.”

*Panel 2: Portrait of Dr. Venom Behind Bars*

Using Midjourney, I attempted to create a realistic portrait of Dr. Venom. The prompt used was “realistic portrait of a single scary green-skinned bald man with red eyes wearing a red coat with shoulder spikes, looking from behind the bars of a prison cell, black background, dramatic green lighting.”

*Panel 3: Angular Ugly Green Spaceship*

The prompt for Midjourney in this panel was “long shot of an angular ugly green spaceship in orbit over a red planet in space in the black sky, dramatic.”

*Panel 4: Advanced Space Fighter Jet Schematic Blueprint*

Midjourney was used again to generate a schematic blueprint of a massive advanced space fighter jet. The prompt was “massive advanced space fighter jet schematic blueprint on a black background, different cross-sections and perspectives, blue streaks and red missiles, star fighter, vic viper gradius.”

*Panel 5: Rectangular Starmap*

This panel showcases a limitation in current AI image tools. Reproducing text correctly in images is still not widely available, and text-to-image does not allow for specific placement or manipulation of elements. In this case, the stars image had to be imported into Photoshop to add the text and lines.

*Panel 6: Reproducing the Iconic Portion*

Reproducing the iconic portion of the image, the three eyes, proved to be a challenge. In this case, in-painting was attempted using Dream Studio, but satisfactory results were not achieved.

*Panel 7: Front-View of the Space Fighter Jet*

Using Midjourney, the prompt for this panel was “front-view of the vic viper space fighter jet on its launch platform, wide wings, black background, blue highlights, red missiles.”

*Panel 8: Close-Up of the Space Pilot’s Eyes*

Again, Midjourney was used to generate a front close-up image of the space pilot’s eyes. The prompt was “front close-up of the black eyes of a space pilot Mr. James Burton peering through the visor of a white helmet, blue lighting, the stars reflected on the glass.”

**Expanding the Canvas with DALL-E Outpainting**

DALL-E’s outpainting tool was used to expand the canvas and fill in the surrounding space with content for one of the images. This workflow requires changing the prompt to describe the portion being crafted at each stage.

**The Progress of AI Image Generation Tools**

It has been a few months since AI image generation tools became widely accessible. During this time, I have had the opportunity to use three of these services: Dream Studio by Stability AI, Stable Diffusion v2.1, and Midjourney.

**Conclusion**

AI image generation tools have immense potential in enhancing old video game graphics. The process of using these tools involves generating multiple images and tweaking prompts to achieve the desired results. While there are limitations and challenges, it is clear that AI image generation tools can create higher-fidelity graphics and breathe new life into old games. With further advancements and improvements, the possibilities are endless.



Leave a Reply

Your email address will not be published. Required fields are marked *

GIPHY App Key not set. Please check settings

The Evolution of UX: A Comprehensive Historical Overview from 1900 to Present | Valerian Kleinschnitz’s Engaging Narrative | July, 2023

Key Takeaways from Leading Product Creation and Management – Keep in Mind the Product