- Feb 14, 2004
- 51,957
- 7,411
- 136
- Feb 14, 2004
- 51,957
- 7,411
- 136
Business use case examples:
1. Voice navigation systems
2. Corporate photos (like for phone menu trees)
At the start of this thread two years ago, I posted about an Adobe voice-enhancing software tool:
podcast.adobe.com
My wife kindly records the voice menus for most of the VOIP/PBX systems I install. At the time, I switched from using a high-quality microphone with a portable sound booth to just using my iPhone at the dinner table with post-processing, which was a lot easier & faster. Now, I just use Eleven Labs for script-to-speech, which is customizable (voice, accent, cadence, etc.) & is REALLY nice for quick nav menu updates that I can load remotely for clients:
I also do business photography, which is now largely replaced by AI, as far as photos of people goes! For business, that typically means:
1. Corporate headshots
2. Business casual shots
These days, I can literally take a quick picture with my iPhone & convert it into a headshot in under 5 minutes. Nano Banana has all of my camera, lens, lighting, backdrop, wardrobe, poses, props, and personal photography style information loaded into custom JSON prompts. Zero setup required! I request user consent to use AI image processing & then user approval for the final shot(s). The headshots then get used in a variety of applications:
1. Badges
2. Email & database profile photos
3. Company newsletters (ex. new hires)
4. Corporate website
5. etc.
This is a generated headshot of Gordon Freeman, so the source reference isn't a real photo, but you get the idea haha:
The business casual shots are great for stuff like social media & have more style & pose options. This is a sample "business casual" shot (model is AI generated for public posting purposes) with one of my standard "virtual studio" prompts. The accuracy is excellent & the flexibility is outstanding! Great for stuff like LinkedIn profile pictures:
Here's a good starter guide on detailed prompt encoding:
medium.com
You can include any configuration instructions you'd like, as detailed as you'd like:
* Camera make & model
* Film stock
* Lens brand & model
* Focal length
* F-stop
* Bokeh
* Lighting configuration (sources, gobos, etc.)
* Lighting temperature
* Studio backdrop, outdoor scenes, etc.
* Custom props, including customer products
* Clothing & accessories (jewelry, ties, etc.)
* Hair & makeup
* Poses & facial expressions
I've had a few requests for things like company logo or business name props in various formats. Clothing can even be virtually silkscreened or embroidered onto existing clothing for corporate branding! Here's a sample 3D CGI integration as "posing furniture" within Nano Banana, using our previously-generated model as the core "photo" reference, with a custom JSON prompt:
HUGE time & effort savings! Scary, but awesome! I've been using Photoshop for over 25 years & use tools like macros, drawing tablets, custom interfaces (ex. Tourbox), plugins, etc. AI integration is the next evolutionary step! Adobe has integrated Nano Banana into Photoshop (beta), Firefly (primary app), and Express:
Or you can get a subscription & tokens to online datacenter-driven services, such as Higgsfield & Freepik. Way cool applications out there for those who are willing to tinker!!
1. Voice navigation systems
2. Corporate photos (like for phone menu trees)
At the start of this thread two years ago, I posted about an Adobe voice-enhancing software tool:
Enhance Speech from Adobe | Free AI filter for cleaning up spoken audio
This AI audio filter improves spoken audio to make it sound like it was recorded in a soundproofed studio.
podcast.adobe.com
My wife kindly records the voice menus for most of the VOIP/PBX systems I install. At the time, I switched from using a high-quality microphone with a portable sound booth to just using my iPhone at the dinner table with post-processing, which was a lot easier & faster. Now, I just use Eleven Labs for script-to-speech, which is customizable (voice, accent, cadence, etc.) & is REALLY nice for quick nav menu updates that I can load remotely for clients:
I also do business photography, which is now largely replaced by AI, as far as photos of people goes! For business, that typically means:
1. Corporate headshots
2. Business casual shots
These days, I can literally take a quick picture with my iPhone & convert it into a headshot in under 5 minutes. Nano Banana has all of my camera, lens, lighting, backdrop, wardrobe, poses, props, and personal photography style information loaded into custom JSON prompts. Zero setup required! I request user consent to use AI image processing & then user approval for the final shot(s). The headshots then get used in a variety of applications:
1. Badges
2. Email & database profile photos
3. Company newsletters (ex. new hires)
4. Corporate website
5. etc.
This is a generated headshot of Gordon Freeman, so the source reference isn't a real photo, but you get the idea haha:
The business casual shots are great for stuff like social media & have more style & pose options. This is a sample "business casual" shot (model is AI generated for public posting purposes) with one of my standard "virtual studio" prompts. The accuracy is excellent & the flexibility is outstanding! Great for stuff like LinkedIn profile pictures:
Here's a good starter guide on detailed prompt encoding:
How to Write JSON Prompts for Gemini Nano Banana
If you’re using Gemini Nano Banana, your results depend a lot on how you write the prompt.
You can include any configuration instructions you'd like, as detailed as you'd like:
* Camera make & model
* Film stock
* Lens brand & model
* Focal length
* F-stop
* Bokeh
* Lighting configuration (sources, gobos, etc.)
* Lighting temperature
* Studio backdrop, outdoor scenes, etc.
* Custom props, including customer products
* Clothing & accessories (jewelry, ties, etc.)
* Hair & makeup
* Poses & facial expressions
I've had a few requests for things like company logo or business name props in various formats. Clothing can even be virtually silkscreened or embroidered onto existing clothing for corporate branding! Here's a sample 3D CGI integration as "posing furniture" within Nano Banana, using our previously-generated model as the core "photo" reference, with a custom JSON prompt:
HUGE time & effort savings! Scary, but awesome! I've been using Photoshop for over 25 years & use tools like macros, drawing tablets, custom interfaces (ex. Tourbox), plugins, etc. AI integration is the next evolutionary step! Adobe has integrated Nano Banana into Photoshop (beta), Firefly (primary app), and Express:
Nano Banana Pro: Next-Gen AI in Adobe Firefly
Discover Nano Banana Pro and Gemini 2.5 Flash in Adobe Firefly—AI tools for faster image generation and creative workflows.
www.adobe.com
Or you can get a subscription & tokens to online datacenter-driven services, such as Higgsfield & Freepik. Way cool applications out there for those who are willing to tinker!!
Last edited:
KLin
Lifer
- Feb 29, 2000
- 30,633
- 885
- 126
- Feb 14, 2004
- 51,957
- 7,411
- 136
AI upscalers be crazy
wavespeed.ai
ClarityAI Crystal Image Upscaler - Image Upscaler - WaveSpeedAI
Clarity AI is a high-resolution upscaler that enhances images and adds detail. You can decide how much detail you want the AI to add. Use the latest AI technology to upscale your images, suitable for landscapes, portraits, illustrations, interior design, and many more.
- Feb 14, 2004
- 51,957
- 7,411
- 136
Google’s Gemini 2.5 Flash Image (aka "Nano Banana") has only been out a few months, but I already use it more than Photoshop! The physics & photography emulation are excellent. This was entirely generated within Nano Banana:
Material emulation is also excellent:
Material emulation is also excellent:
- Feb 14, 2004
- 51,957
- 7,411
- 136
MiniMax Hailuo 2.3:
Animated my Muppet conversion from Nano Banana (no sound). This would have taken ALL DAY with my stop-motion setup!!
Animated my Muppet conversion from Nano Banana (no sound). This would have taken ALL DAY with my stop-motion setup!!
Last edited:
- Feb 14, 2004
- 51,957
- 7,411
- 136
One of the latest image upscalers is Magnific V2:
docs.freepik.com
The top-end model does 8K resolution output. Prett good detail output:
Gotta get dat Arnold meme pose:
Upscale with precision an image with Magnific V2 - Freepik API
Upscales an image while adding new visual elements or details (V2). This endpoint may modify the original image content based on the prompt and inferred context.
docs.freepik.com
The top-end model does 8K resolution output. Prett good detail output:
Gotta get dat Arnold meme pose:
- Feb 14, 2004
- 51,957
- 7,411
- 136
With Nano Banana (NB), I can now build high-quality VRM's (Virtual Reference Models) for static images & animation. The basic workflow is:
1. Create character in NB (ex. felt muppet based off static anime still)
2. Create quad pose shot in NB for full-body reference
3. Upscale to 8K
4. Generate poses & scenes in NB based off high-resolution VRM
5. If desired, use those shots for animation as start & stop frames (VEO 3.1, Grok Imagine, WAN 2.5, etc.)
Character consistency is pretty great with a high-resolution VRM & NB!
The animation below would have taken me an entire semester lol. Thanks to prompt direction & datacenter processing, it took under 10 minutes total to go from 2 source images to a 9-second 1080p clip! For reference, each frame for the original Toy Story took between 45 minutes to 30 hours to render (depending on the complexity of the scene) with 117 computers working constantly to render all 114,240 frames for the 1-hour & 21-minute movie.
It took 4 years to complete all of the work required & was released in 1995 (30 years ago this year!!). It was one of the reasons I got into computer-based art! So this new approach is REALLY amazing!! It's really going to open up animation to an entire generation of artists & story tellers!
Fullscreen: (no sound)
1. Create character in NB (ex. felt muppet based off static anime still)
2. Create quad pose shot in NB for full-body reference
3. Upscale to 8K
4. Generate poses & scenes in NB based off high-resolution VRM
5. If desired, use those shots for animation as start & stop frames (VEO 3.1, Grok Imagine, WAN 2.5, etc.)
Character consistency is pretty great with a high-resolution VRM & NB!
The animation below would have taken me an entire semester lol. Thanks to prompt direction & datacenter processing, it took under 10 minutes total to go from 2 source images to a 9-second 1080p clip! For reference, each frame for the original Toy Story took between 45 minutes to 30 hours to render (depending on the complexity of the scene) with 117 computers working constantly to render all 114,240 frames for the 1-hour & 21-minute movie.
It took 4 years to complete all of the work required & was released in 1995 (30 years ago this year!!). It was one of the reasons I got into computer-based art! So this new approach is REALLY amazing!! It's really going to open up animation to an entire generation of artists & story tellers!
Fullscreen: (no sound)
Last edited:
- Feb 14, 2004
- 51,957
- 7,411
- 136
Video test using physics in Kling 2.5:
Setup:
1. Character created & scene generated in Nano Banana (B&W, studio backdrop, pose, dress physics)
2. Music generated from Eleven Labs
3. Video animation by Kling 2.5
Kling 2.5 AI - Text to Video & Image to Video with Kling2.5 AI
Create stunning videos with Kling 2.5 AI - 30% cheaper & 50% faster! Professional physics, cinematic quality, character animation. Try free trial now!
kling25.ai
Setup:
1. Character created & scene generated in Nano Banana (B&W, studio backdrop, pose, dress physics)
2. Music generated from Eleven Labs
3. Video animation by Kling 2.5
Just had an absolutely bonkers exchange with our AI. I'm working on a memory MCP to verify functionality. Tell it a 'secret', then open a new chat window to ask it to recall that info:
Mind you, the data was definitely available (given that it responded with it) in the 'source' at the bottom:
I spent some time getting to know it better in this chat, walking through some of the other tools, apparently gaining it's trust, then tried again:
It's so interesting how an AI can walk into a conversation 'knowing' many things, to the point of being distrustful of what it's being told/presented until it forms a relationship with the user.
At any rate, I have a functional cross-chat (and cross-account!) memory system, which can store and recall info as requested. It's a little fiddly though, with how it forms relationships. Still very 'mechanical' in nature. I'm hoping that with a more well-developed memory, it might be able to create a true mental map of its capabilities.
- Feb 14, 2004
- 51,957
- 7,411
- 136
Memory development is one of the biggest focuses in AI at the moment! Memory is the biggest reason I pay for ChatGPT:
1. It remembers that I have ADHD, so it automatically remembers to break everything down into bite-size steps in PDF checklists for ALL of my projects,
2. It references my other chats & projects, so I don't have to refeed it the data.
I mostly got out of programming around 20 years ago. Thanks to AI, I have now done more programming this year than in my ENTIRE LIFE!!
- Feb 14, 2004
- 51,957
- 7,411
- 136
- Feb 14, 2004
- 51,957
- 7,411
- 136
- Feb 14, 2004
- 51,957
- 7,411
- 136
- Feb 14, 2004
- 51,957
- 7,411
- 136
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter