Photoshoot Catalog Creation - claid.ai - Startup Analysis

claid.ai

Key Vision Lessons

• Product placements with coordinates guided
• Image operations - Resizing, restorations, color adjustments, padding, super resolutions
• Detailed examples for retail, real estate

Keep Exploring!!!

AI Products ?

Before embarking on any ambitious AI work, ask yourself these questions:

• How much time do you plan to spend?
• Do you plan to customize or reuse APIs?
• Do you have data to validate consumer needs?
• Are you building the product with the end customer in mind or working with a matrix of stakeholders?
• How do the realistic technical skills compare to the gaps in creating actual products?
• Many developers create "Hello World" apps that masquerade as real-world solutions, but in reality, only truly practical applications survive.

Without data, without product clarity, and without customer collaboration, it will end up a big success :)

#2024 will have many #GenAI apps in market. Many "Hello World" LLM apps masquerade as real-world solutions. Without data, without product clarity, and without customer collaboration, it will end up a big success :) that is the going to be the mantra of #GenAI adoption. 

Keep exploring!

Keep Exploring and Move on

Ideas are easy, On ground challenges makes the differences
Power point is easy, Getting first principles right is important
Always strong basics and persistence shine over presentations 
Ideation is Easy, Evolving is Innovation and Differentiator

Keep Exploring!!!

Startup Analysis - Jan20 - Concepting Tools in Vision - dreamlook.ai

Concepting, Vision based ideation is picking up. Dreambooth custom fine-tuning is easy to use, intuitive, and user-friendly. The UX and execution are seamless. 

Below are steps in a basic working example

1. Start with Custom Model Training

2. Upload Images

3. Submit a Job

4. Monitor Job Progress

5. Job Completion

6. Generate Images based on Custom Models

Very user friendly tool.

Keep Exploring!!!

Startup Analysis - Jan 29 - yarnit.app - Concepting Tool

Concepting Tool 

Creative use of LLM, GenAI, and Vision Models

• Ideate, Design, Write, Audit & Publish content 
• 50+ expert-trained templates
• Contextual content ideas 
• Vision Tools - Dreambrush, background remover

Keep Exploring!!!

“Can ML solve my problem?

What it takes to get to the level, ML Perspectives

Even answering the question “Can ML solve my problem?” requires you to overcome half of the challenges ML libraries, Find state-of-the-art (SOTA) deep neural networks, experiments, and MLOps.

Machine learning often boils down to the art of developing an intuition for where something went wrong.

Certain behavior signals with where the problem likely is in your debugging space - preprocessing, data issues, optimization, weak labels, and learning rates. ML Learning = Experimentation = Experience.

The efforts in terms of marking production-ready apps - Text, Code copilot are in (primetime)

Vision tools/custom tools are evolving rapidly. The maturity of imagegen, DALLE3, and Midjourney is nearing prime time in 2024, One area is learning to build production solutions where there is maturity. Another area is building custom tools where maturity is nearing prime time :)

Ref - Link1, Link2

Keep Exploring!!!

Figma - Lessons we can learn on Usage / Adoption / ML Lens

How to use user signup data

• Email, name, and role
• Data shows us how the product is used, and includes metadata about how the platform is accessed
• Features our users are using
• Features like invite other users into their file, to manage file permissions, and to publish their work
• Primary drivers for crashes to improvements incremental frame loading and image sampling for prototypes

How to know user engagement

• Funnel with all users at the top
• Encourage users to interact with notifications
• User behavior at scale
• Those on personal accounts, designers on team accounts were opening 80% of the comment notifications

Data Science Perspectives

• Significant lift in all users leaving comments, and this increase was especially pronounced for non-designers
• Data also indicated a more dynamic collaboration process

Link1, Link2, Link3

Keep Exploring!!!

2024 Tech Goals

• Adapt to a flexible solution mindset, More tools will come :)
• Bird's eye view of offerings / Tools
• Experiment with Tech, Build working solutions
• ML / Cloud solution architectures across clouds
• Teach with use cases/solution - Product + Developer Perspective
• Apply for a few accelerators for solutions
• Build working solutions
• AI for a social cause
• Solutions E2E integrating aspects

Keep Exploring!!!

Buying Decision - Data Analysis

This could be biased but when you have limited budget and have to take a convincing decision :)

 

Keep Exploring!!!

Good Read - Staying up With Experience in Tech

Summarizing key points from post

• Have solution / Approach / Code on key areas/ New Tech
• Spend Time on Architecture / Review / Discussions
• Understanding the fundamentals of the products you use
• Map Tech trends to use cases
• 'connecting the dots' = Product + Tech + Domain

Keep Exploring!!!

Fantastic Read on Titles vs Learning vs Relevance

Fantastic Read on Titles vs Learning vs Relevance 

Key pointers

• Irrelevance is the new retirement
• As humans we struggle for relevance
• Growing old is compulsory but growing up seems optional
• The pain of failure is less than the pain of regret
• You will be relevant in job market as long as you keep your learning curve focused

Keep Exploring!!!

Leadership vs The Job Security Myths Vs Layoffs

Key Article from Googler

• Pattern #1 - They point in a direction, their subordinates swarm the area, try a bunch of stuff, and sometimes something sticks and is cool.
• Pattern #2 - Given that they have no real vision of their own, they really need their subordinates to come up with cool stuff for them
• Pattern #3 - Just randomly firing people, torching institutional knowledge, and blowing up perfectly functional teams.

So I guess I will just hang around and do my job until Google no longer wants me.

Job Security Myths

There is no relationship between skills vs job vs priority. We need to carve our own skills to survive.

• When you know tech, you feel like know business to remain competitive
• When you know Database, you feel like learning API to remain competitive
• When you know ML, you feel like learning Kubernetes, and MLOps to remain competitive
• When you know UI, you feel like learning DB, and API to remain competitive

Add everything that compliments, You can only be a better version every day not by comparison but by your own aspirations

Keep Exploring!!!

2024 - GenAI Trends - Use Cases

How Enterprise Companies are Buying AI (or Not) with ContextualAI, Anthropic, and Glean

Use Cases

• Info Discovery and Synthesis
• Deeper Insights
• Hierarchical Summarization
• Support Chatbots
• Knowledge Extraction

Barriers to Adoption

• One tool is better than the other
• Security questions / Data Leaks
• Governance to manage tools/data

Challenges

• Tech does not work from Day 1
• It needs iterations
• Fixing Hallucinations / Citations
• Focus on use case vs Fine tune vs Level Setting vs Context Window 
• Artificial specialized intelligence = Fine tune vs Context Window 

Product Roadmap

• Getting Certifications
• Run Lean in Customer Environment
• Solution = LLM + RAG + VectorDB - Blend of All (Solution Strategy)

Keep Exploring!!!

AI Tools + Vision Use Cases + GenAI

Vision Tools + GenAI

• Stable Diffusion, ComfyUI and Automatic1111.
• Dreambooth and LoRA
• Midjourney, Dalle, Runway, and PikaLabs
• Supportive AI tools for segmentation, data labelling and inspection
• NeRFs and Gaussian Splatting
• DALL-E, Runway and Wonder Studio

Use Cases

• Commercial Production
• Graphic Design
• Social Media
• Content Marketing
• Branding
• Product Mockups
• Spec Ads

Domain-Specific Use Cases

• Drafting concept art, architectural concepts, and interior design plans on a budget
• Generating free portraits of yourself, friends, family members, and pets
• Completing hand-drawn projects that you no longer have free time for
• Designing stunning cover art for podcasts, albums, and books
• Printing AI-generated posters that fit your aesthetic
• Crafting custom gifts for birthdays and holidays
• Generating wallpapers and backgrounds for your desktop or phone
• Visualizing random ideas to get your creativity flowing
• Mixing up your social media posts with a new style
• Writing cards and invitations for personal and commercial use
• Creating eye-catching clipart-style characters for emails, posts, and presentations
• Developing logos and icons for websites, apps, and marketing
• Experimenting with fashion design projects
• Competing in art challenges to embrace the AI community
• Growing your business with AI art prints

Keep Exploring!!!

Virtual Try-on (VTON) Research and Startups

Startups in Focus

• Try on Clothes -  Vmake.ai, virtulook wondershare, Alibaba's AI Magic, zero10.ar, 3dlook.ai
• Virtual Try on Beauty- Perfect365, Orbo
• Virtual Try on Jewelry - Tangiblee Virtual Try-On
• Image Catalog Management - Vue.ai, Spyne.ai, Stylitics, wide eyes
• Face Beautifier - Banuba

Key papers Repos

• A Curated List of Awesome Virtual Try-on (VTON) Research
• Paint-by-Example
• DCI-VTON-Virtual-Try-On
• StableVITON: Learning Semantic Correspondence with Latent Diffusion Model for Virtual Try-On
• FCN, U-Net models implementation in TensorFlow for fashion clothing parsing
• Clothing Matching & Alignment (2D)

Code Examples

• Deforum_Stable_Diffusion
• CP-VITON-PLUS.ipynb
• Viton_hd.ipynb

Image Editing

• Poisson-Image-Editing
• Laplacian-Pyramids
• BlendModes
• Poisson image editing
• ImageStitching
• blend_modes
• Awesome-Image-Composition
• Image Blending Using Laplacian Pyramids

Keep Exploring!!!

Comfy Tool Notes

Comfy Tool Notes

Summary from Link

Key Notes

• Model files - civitai, hugging face
• CLIP, Main Model, VAE
• CheckpointLoader - Outputs Model, Clip, VAE
• Clip Model - Encode the text to main model, Positive and Negative prompt
• Encoded positive and Negative prompts sent to MODEL at each step and used to guide denoising
• VAE transalate image in latent space to pixel space

Inpaint Examples

Samplername - uni_pc_bh2

• AutocodePro
• Finetuned Stable Diffusion for Anime
• AlphaCTR
• Low Rank Optimization LoRA models are essentially compact versions of Stable Diffusion that introduce minor, yet impactful modifications to the standard models. 
• ControlNet/T2I adapter needs the image that is passed to it to be in a specific format like depthmaps
• Stable Zero123 is a diffusion model that given an image with an object and a simple background can generate images of that object from different angles.
• SDXL Turbo is a SDXL model that can generate consistent images in a single step. 

Nodes Explanation

• CLIP model: to convert text into a format the Unet can understand
• Unet: to perform the "diffusion" process, the step-by-step processing of images that we call generation
• VAE: to decode the image from latent space into pixel space (also used to encode a regular image from pixel space to latent space when we are doing img2img)
• KSampler node. This is the actual "generation" part, so you'll notice the KSampler takes the most time to run when you queue a prompt.

Checkpoints

• Place checkpoints in the folder ComfyUI/models/checkpoints:
• SDXL 1.0 base checkpoint, SDXL 1.0 refiner checkpoint
• VAE - Place VAEs in the folder ComfyUI/models/vae
• Fixed SDXL 0.9 VAE 
• LoRAs - Place LoRAs in the folder ComfyUI/models/loras
• Stable Diffusion Hub

Keep Exploring!!!

AI threats, freelancing, making millions & building businesses

AI threats, freelancing, making millions & building businesses

Good talk and lot of key pointers. Execution matters + creative ideas.

Key Ideas to Align

• Time to execution is low
• Top creators can do more work 
• Things can be done in parallel
• Weight of brand carries message (AI or AI-generated)
• Increasing opportunities for a few
• Embrace AI for better execution
• Pulling tools together makes you more competitive
• Ideas are connecting the dots

AI will take over

• Content writing
• The cost will come down to building software
• You can better learn task / prompts with chatgpt
• Better Experience = Better Prompt
• Good prompt based on past experience
• Drag and Drop pics train vision model

Tips

• Get more Inspirations
• Add more ideas
• Synthesize tools (Good/bad output)

AI Risks

• Not works on consumer side
• Don't do things already mainstream

Path to Progress

• Freelancer -> Agency -> SaaS Platform
• Posters for Real estate
• Posters for Mfg Companies
• Niche Photography
• Upskill before it becomes popular
• Storytelling, Personalization

Future

• Multimodals
• AI + Image + Vision
• Super Apps
• AR, VR
• Keep good at some skills
• Skills + Network + Opportunities
• Identify gap in market to graph your career :)

Keep Exploring!!!

Vertex AI Search

Example for Vertex AI Search and Conversation. In GCP look for Vertex AI Search and Conversation

• We will upload PDFs
• Ingest them and perform qna with it

Step 1- Select Search Option 

Step 2 - Name your App

Step 3 - Create Datastore for APP

Step 4 - Upload pdf to the App

Step 5 - Configure your data store

Step 6 - Configure and Customize

This tutorial was helpful to follow/evaluate. Link

Keep Exploring!!!

GenAI based AI interior tools

Very interesting prototype and design choices from post

Key Learning's

• 360 view of the room
• GenAI based image inpainting
• By using a 360° photo of the space we can move around in different positions and sizes to have a preview of each section, and we can fully expand the space to have a 360° immersive experience as shown below.
• Generative AI is inpainting and outpainting
• Regenerate variants to view more options

Keep Exploring!!!

Reasons for change of Role / Job Switch

• Some roles are toothless, You do the same thing over the years
• Your motivation will get killed when your ideas get killed
• Working with clients who do not see the big picture will not help you grow
• Life is short when you are @ 40, You cannot wait for 5 years with the hope that you will get to your goals
• The world is full of busy people implementing their ideas, When you are not in a position to implement your idea you need to find a place where you implement/own/improve your idea
• Demos get demolished after the meeting; Real value / Innovation building is a continuous effort not a sprint job

Keep Exploring!!!

Vision - Duplicate Images

Keep Exploring!!!

Vision Learning Startup - curiousrefuge

curiousrefuge

Made an Adidas AI Spec Commercial during Coffee Break
byu/Theblasian35 inmidjourney

The World’s First Home for AI Storytellers 

Ai-filmmaking

• Ideation + Scriptwriting
• Art Direction + Curation
• Prompt Mastering + Directing
• Pitching + Storyboarding
• Editing + Pacing + Character
• Cinematography + VFX

AI Filmmaking Tools

The Best Text-to-Image AI Tool

• Midjourney
• Adobe Firefly
• Dall-E 3
• Leonardo
• Stable Diffusion

The Best Image-to-Video AI Tool

• Runway Gen 2
• Pika Labs

Text-to-Video AI Tool

• Runway Gen 2
• Moonvalley

Language Processing: Script Development, Outlining, Research, Distribution, & More. ChatGPT4

The Best AI Music for Filmmakers Tool

• Soundful
• Stable Audio 
• Google MusicML 

AI Text-to-Voice Tool for Filmmakers - Elevenlabs

Best AI Tool for Voice Cloning - Elevenlabs

Best AI Tool for Generative Inpainting - Midjourney Inpainting

• Demo Videos
• Midjourney Inpainting Tutorial
• OpenAI’s AI Filmmaking Research | The Curious Refuge Podcast

Keep Exploring!!!

LLM Discussions - Good Read

Unfortunately , too few people understand the distinction between memorization and understanding. It's not some lofty question like "does the system have an internal world model?", it's a very pragmatic behavior distinction: "is the system capable of broad generalization, or is… https://t.co/1fagV1YI15

— François Chollet (@fchollet) December 15, 2023

Future LLMs will be a different breed than current LLMs, so what's true for current LLMs might not apply in 1 year:

- Self-play: Allows these models to learn beyond human data, like AlphaGo

- Bootstrap: if validation is easier than generation, the models can bootstrap by…

— Will Bryk (@WilliamBryk) December 24, 2023

Unfortunately , too few people understand the distinction between memorization and understanding. It's not some lofty question like "does the system have an internal world model?", it's a very pragmatic behavior distinction: "is the system capable of broad generalization, or is… https://t.co/1fagV1YI15

— François Chollet (@fchollet) December 15, 2023

Limits of Transformers on Compositionality

• First, transformers solve compositional tasks by reducing multi-step compositional reasoning into linearized path matching. 
• This contrasts with the systematic multi-step reasoning approach that learns to apply underlying computational rules required for building correct answers [71, 37, 27]. 
• Shortcut learning [29] via pattern-matching may yield fast correct answers when similar compositional patterns are available during training but does not allow for robust generalization to uncommon or complex examples. 
• Second, due to error propagation, transformers may have inherent limitations on solving high-complexity compositional tasks that exhibit novel patterns. Errors in the early stages of the computational process can lead to substantial compounding errors in subsequent steps, preventing models from finding correct solutions.

Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solve

• Humans and large language models (LLMs) have some shared properties and some properties that differ. If LLMs are analyzed using tests designed for humans, we risk identifying only the shared properties, missing the properties that are unique to LLMs (the dotted region of the diagram). We argue that to identify the properties in the dotted region we must approach
• LLMs on their own terms by considering the problem that they were trained to solve: next-word prediction over Internet text.

On the Measure of Intelligence

• Describing intelligence as skill-acquisition efficiency and highlighting the concepts of scope, generalization difficulty, priors, and experience, as critical pieces to be accounted for in characterizing intelligent systems.
• Intelligence as a collection of task-specific skills
• Intelligence as a general learning ability
• Skill-based, narrow AI evaluation
• The spectrum of generalization: robustness, flexibility, generality
• System-centric generalization: this is the ability of a learning system to handle situations it has not itself encountered before. 
• Developer-aware generalization: this is the ability of a system, either learning or static, to handle situations that neither the system nor the developer of the system have encountered before.
• Local generalization, or “robustness”: This is the ability of a system to handle new points from a known distribution for a single task or a well-scoped set of known tasks
• Broad generalization, or “flexibility”: This is the ability of a system to handle a broad category of tasks and environments without further human intervention

Keep Exploring!!!