🙄AI Ads need to chill

newslounge.co · March 15, 2024

Good Morning. 

This is newslounge, We’re like your morning coffee - essential, energizing, and slightly addictive.

In a recent interview, OpenAI's CTO, Mira Murati, hinted at Sora's release within a few months, contrasting earlier statements from Sora's tech leads who gave no definitive timeline, suggesting there might be some communication or planning discrepancies.

On Today’s menu:

  • Netflix and Apple at the Oscars

  • Underwater World - Case Study

  • LoRA Training in Kohya_ss

  • Midjourney's New "Character Consistency"

  • Stable Diffusion Integration for Nuke

  • AI Ad Under Scrutiny

-Ardy

was this email forwarded to you? you can sign up here.

🚨Just a quick note: You can share this newsletter with your friends using the unique referral link at the bottom of this email. By doing so, you'll earn points that can be exchanged for some cool prizes. Let's spread the word! 😎

HEADLINE

A ROLLERCOASTER RIDE FOR NETFLIX AND APPLE AT THE OSCARS

While streaming dominates the entertainment industry, it didn't translate to success at the Oscars. Streaming platforms' nominations saw minimal wins.
Netflix secured only 1 out of 19, and Apple didn't win any of its 13 nominations.
At the 2024 Oscars, streaming platforms had a tough night. Netflix managed only one win out of 32 nominations for both Netflix and Apple combined, with Apple failing to win any awards.
Netflix's single victory came with Wes Anderson’s "The Wonderful Story of Henry Sugar," winning the live action short film category. This film, which features stars like Benedict Cumberbatch and Dev Patel, marked Anderson's first Oscar win, despite him not being present to accept the award.

Netflix has been accumulating Oscars since 2017, reaching a total of 23. However, it has yet to win in the best picture, lead actor, or actress categories. This year, notable losses included "Maestro" for best picture and nominations in acting categories for films like "Maestro" and "Rustin."

Martin Scorsese, with a nomination for "Killers of the Flower Moon," has won only one Oscar previously, in 2007 for "The Departed." His film "The Irishman" also didn’t win any Oscars in 2020 despite multiple nominations.

Netflix’s Oscar wins over the years include various categories, from directing ("The Power of the Dog" by Jane Campion) to documentary features like "Icarus" and "The White Helmets." In 2019, "Roma" by Alfonso Cuarón won three Academy Awards. Amazon Studios also saw success in 2017 with three Oscars for "Manchester by the Sea" and "The Salesman."

-NL

CASE STUDY

SUBAQUATIC WORLD - HOUDINI & COMFYUI

Here's a brief reminder for our new readers interested in learning more about this project and its workflow. Below, you'll find all the necessary links.

One more thing:
Please use the poll section below to share your feedback on this case study.

🚀 Youtube Tutorial: here
🎬 Youtube Promo: here
👩‍💻 Project Breakdown here
🎁 Download the Workflow here

WHAT’S NEXT?

credit: Ardy / LoRA SD15 of Kaya Scodelario

I'm working on another realistic Houdini & ComfyUI animation featuring a 1969 Corvette 3D model. My goal is to enhance the appearance of my car model, with a special focus on the interiors, by employing a refined LoRA.
To assess the efficacy of LoRA, I conducted an experiment using 25 of Kaya Scodelario’s photos to train and test various performance and accuracy settings.
After almost 3 days of playing with Kohya_ss, I was finally able to get a decent result (and go from the left image to the right image).

🤿Next, I plan to incorporate the same JSON to train the Corvette LoRA.
This photo was achieved using EpicRealism Checkpoint + my KayaLoRA
🚨Stay tuned for a full rundown on this project plus more info on training interfaces and solutions.

INTERESTED TO LEARN MORE?

Login or Subscribe to participate in polls.

💡What is a LoRA?
A LoRA is a method to achieve better quality in characters, poses, scenes, or concepts beyond what the base model can provide. The base model is foundational for generating images, containing all the trained knowledge about various objects and beings, derived from billions of images and text. However, it might not cover everything in detail, leading to imperfect outputs.

LoRAs are essentially like targeted training sessions, allowing you to enhance the base model's output on specific subjects that aren't well represented or detailed in the original training. What's incredible about LoRA is its adaptability to concepts/styles, making it flexible to combine with the base model.

For example, you could train a LoRA on an anime character to capture their appearance and style, and then apply this LoRA with a realistic base model to create a realistic rendition of that character.

-Ardy

🙋‍♂️WE NEED YOUR FEEDBACK…

DID YOU FIND THIS 'CASE STUDY' HELPFUL?

I'd love to hear your thoughts.

Login or Subscribe to participate in polls.

We are seeking:
full-stack software engineer
web developer

SNIPPETS

😲Meta has accused an ex-vice president of stealing proprietary AI intel, including info on employees who later joined the VP's startup.

🤖Google won’t let Gemini, its AI chatbot, answer election-related questions in any country where one is taking place.

👩‍💻Google is changing how Chrome detects and warns you about unsafe sites.

📈Reddit plans to raise up to $748m through its forthcoming IPO, with ~22m shares priced between $31 and $34 each.

🤿'Disney Adult' industrial complex, Was it intentional?

🚀Elon Musk now has another reason to hate OpenAI. A startup called Figure, which has teamed up with OpenAI, recently showcased a humanoid robot that appears to be more advanced than anything demonstrated by Tesla's Optimus robot, according to a video they released on Wednesday.

AROUND THE WORLD OF AI

EXCITING UPDATE IN MIDJOURNEY

Credit: Luis Di Lascio

👩‍💻Okay, my friends, the wait is finally over! The community's top-requested MidJourney feature, “Character Reference,” for character consistency, is finally HERE!
Here’s how to use it.
Generate an image and copy the image URL.
Now all you have to do is add the —cref +url command of the character to your Prompt.
And with —cw you can specify the weight of the reference —cw 0 will catch only the face, but with —cw 100 also the clothes and accessories.

💡Quick Thoughts:

  • It's not perfect...but it's close

  • Be detailed in your prompt for clothes/features

  • Use this on characters made from Midjourney images.

  • Works similarly to image prompts

  • Except it 'focuses' on the character "traits"

  • Precision is limited,

  • It won't copy exact dimples / freckles / tattoos/ etc

  • Works on regular and Niji models.

-NL

STABLE DIFFUSION INTEGRATION FOR NUKE

Danilo de Lúcio

💡NukeDiffusion, developed by Danilo De Lucio, is a powerful integration tool for Nuke, leveraging Stable Diffusion for generating AI images directly from textual prompts using local checkpoints. This solution utilizes the official Hugging Face library and operates entirely locally, eliminating the need for account creation.

🚀For detailed workflow and installation guidelines, visit its GitHub page.
▶ Watch it on youtube here

-NL

AI AD UNDER SCRUTINY

Under Armour x Anthony Joshua

A new ad from Under Armour featuring boxer Anthony Joshua has come under fire from creatives on Instagram after its director claimed it as the “first Ai-powered sports commercial” — but critics in the industry say it blatantly reused others’ work without credit as part of an AI hype cycle cash grab.

Director Wes Walker posted the spot, along with several variations and riffs, on Instagram earlier this week, saying: “Under Armour asked us to build a film from nothing but existing assets, a 3D model of Anthony Joshua and no athlete access. This piece combines Ai video, Ai photo, 3D CGI, 2D VFX, Motion graphics, 35mm film, digital video and advances in Ai voiceover. Every current Ai tool was explored and pushed to the maximum.” [I have left “AI” as “Ai” throughout.]

🤷‍♂️The crazy thing is, in my opinion, it was a lazy approach, and it doesn't even look that good! Moreover, they didn't credit the footage used, which was shot 2 years prior for another Under Armour ad directed by Gustav Johansson. A professional courtesy that costs nothing and would have more honestly represented who actually created the images seen here.

-Ardy

😂That’s it for today, I’ll leave you with this!!

OpenAI's CTO, Mira Murati

-newslounge

What'd you think of today's edition?

Login or Subscribe to participate in polls.

BUT, WAIT! THAT’S NOT ALL! ⬇

🎁You can get free stuff for referring your friends!!

Earn free gifts 🎁

5 referrals - “Water Bottle Stickers” 🔺15 referrals - “Mystery Box”
25 referrals - “Water Bottle” 🔺 40 referrals - “Nuphy Desk Mat”
60 referrals - “Logitech Mouse”