Spencer

official everything designer of the twitter health space™
messing with AI every now and then

[Spencer @rendererr]

Generate AI slop so good that normies won't hate on you in your Meta Ads comments

You're looking at 10 minutes of work across all 3 images

Drop your brand and I'll make you a product image using my new system:

[Spencer @rendererr]

Here's my system that generates unlimited editorial-quality product images with AI

(Takes 3-5 mins per usable image)

It covers how to:

•Ideate tasteful ideas
•Generate high-quality clean product shots
•Preserve product accuracy 95%+
•Do it all over and over again

https://t.co/WyPvLARc7F

[Spencer @rendererr]

Fully AI generated using the new system

Second one is a little off-brand for you but I thought it looked cool anyways

[The Crypto Dog 📈 @TheCryptoDog]

@rendererr https://t.co/HvuzWmjk5B

[Spencer @rendererr]

@biohacker 💯

[Spencer @rendererr]

If you use AI for anything, you need to be aware of a psychological phenomenon called Gell-Mann Amnesia.

When you use AI for something you understand intimately, you'll notice missing details or inaccuracies that would clearly mislead anyone who doesn't know better.

However, when using THE SAME models for something you're less familiar with, you'll watch it output a seemingly correct answer with sources cited, and you subconsciously trust it as the full, accurate truth.

Now that models are getting better, blatant hallucinations are less common, but in most fields there's usually a MASSIVE gap between what you can learn from somebody who's tried everything, and the general birds-eye, broad knowledge you can get from asking AI.

I tried replicating my own product image generation system with Opus 4.8 as this guy suggested, and the instructions it produced advised me to do things that I specifically warn against (because I've already tried them a million times and they don't work).

If you have unlimited time and resources, you probably can figure it all out yourself, or you can pay a little and get instant results that are the culmination of 100+ hours of my own trial and error.

[Hins Liu @EcomHins]

@rendererr to everyone reading this, just ask Claude how to do it, copy paste this tweet and its images in and the prompts it gives you get it done 1:1

[Spencer @rendererr]

Generate AI slop so good that normies won't hate on you in your Meta Ads comments

You're looking at 10 minutes of work across all 3 images

Drop your brand and I'll make you a product image using my new system:

[Spencer @rendererr]

@EcomHins Oh right why didnt i think of that

[Spencer @rendererr]

Grayhat dropshippers run AI slop meta ads. Maybe that's fine if your ICP are pajama people who eat frozen food with their shoes on inside.

But if you run a halfway serious brand, you shouldn't do that.

I generated these pics within minutes of each other and it cost me almost nothing.

This system genuinely outputs pics at this quality OVER AND OVER.

We're talking
•95%+ product resemblance
•Consistent photoshoot-quality output
•3-5 minutes of effort per usable image

Putting a link to the strategy in the replies

[Spencer @rendererr]

+ a lot of the ads I see from large established companies don't even have their own product in the right size/proportions, and the most important text on the product label are all messed up (logo + product name)

This system does not promise 100% perfection but it puts you leagues above everything out there short of real photoshoots

[Spencer @rendererr]

I see brands owned by fortune 500 companies that are running super sloppy AI ads all the time- far far worse than this; both in text accuracy and aesthetically

Meta Ads demands LOTS of creatives now. The strategy has shifted

AI blemishes are more or less accepted as part of ecom in general now. Consumers don't care to any meaningful extent especially if they don't even notice

[Spencer @rendererr]

@ashmit__rana Lfg

[Spencer @rendererr]

@mweinbach bro is a lake drainer

[Spencer @rendererr]

I would not count on 100% accuracy every time, but that's a model limitation.

The purpose of the system is to save you $$$ while giving you near-unlimited high quality brand assets for meta ads, email marketing, and landing pages.

If pixel-perfect accuracy every time is a requirement, this probably isn't for you

[Spencer @rendererr]

Yes, right now it's a 30 minute video about how to use Claude and Higgsfield together. It focuses on 4 main pillars:

•creative strategy/ideas
•consistent product resemblance
•getting an editorial look (avoiding sterile AI vibes)
•systemizing everything so you can make images FAST

I am also offering free 1:1 consultations to everyone who buys right now in case you need help or can't get good results. Customer feedback will help me make it better.

Price is currently $195 with the discount code but as I add more to it, price will probably increase. Anyone who buys at the current price will get ALL future updates for free.

Also just saw your DM - here's that pic for you from the other post. Made this with the exact system.

[Spencer @rendererr]

Here’s the system:

Use code TWITTER for a large discount (Ending in 48 hours)

https://t.co/ghpV1hVmOo

[Spencer @rendererr]

Great Cowork use case for autists: Clean/organize your desktop.

Just getting 90% of it sorted into folders takes away the burden of looking at the whole mess

Any extra screenshots or files go to an archive folder for you to check before deleting

Before & after:

[Spencer @rendererr]

@DTCWineGuy Done

[Spencer @rendererr]

Generate AI slop so good that normies won't hate on you in your Meta Ads comments

You're looking at 10 minutes of work across all 3 images

Drop your brand and I'll make you a product image using my new system:

[Spencer @rendererr]

@jxnlco Can get pretty far with just Claude and Higgsfield

[Spencer @rendererr]

@Stelios2679 ?

[Spencer @rendererr]

@trowalebarton 🫡

[Spencer @rendererr]

Just dropped a course on how to create a system that generates high quality product images with AI. It covers how to:

•Quickly create high-quality clean product shots
•Preserve product accuracy/fidelity
•Do it all over and over again

https://t.co/ghpV1hVmOo