Systematiq Ai

MINIMAX OUTTAKES V2 → ABSURD I swear it has comedic timing. It’s Saturday…so we’ll keep it light. **This is purely for entertainment** MIDJOURNEY PROMPT: movie still of a comedy film that was never made, and movie we never needed lol --chaos 10 --ar 16:9 --p--stylize 200 --v 6.1 MINIMAX PROMPT STRUCTURE: The camera [motion] as 2 men [action] MINIMAX PROMPTS: 01. The camera follows as one sprays the other man with a hose, comedy movie 02. The camera slowly pushes in as 2 men begin to ballroom dance with eachother 03. The camera follows as an ostrich walks into the scene. One man pets the ostrich awkwardly 04. The camera follows as 2 men run to the right as they are chased by a goofy t-rex 05. The camera pans down as two men crouch to pet an alligator 06. The camera follows quicky as two men cannon ball into a kiddie pool 07. 2 men being sprayed by a fire extinguisher 08. The camera slowly pushes in as two men play the air-guitar 09. The camera whip-pans right to reveal a goofy mascot doing an awkward dance The camera follows as frankenstein walks into the frame and waves at the camera awkwardly QUICK THOUGHTS: Minimax has a different element It lacks certain controls But it has creative element to it The vintage aesthetic tends to work better (for me) It feels like the only (video) tool  That comprehends humor Try prompting for ridiculous outcomes. It definitely helps the learning curve. Have fun. PS: For more weird content…join the party and sub to the newsletter → LINK IN COMMENTS #minimax #runwayml #midjourney #ai #innovation 

Midjourney → New External Editor Feature. I have some thoughts…(not all good) Took it for a test-drive on the Tesla Robotaxi. To give it a facelift. WHAT IT IS: Edit external images in MJ Or edit MJ generated images YOU CAN ACCESS IF YOU HAVE: A yearly membership Been a subscriber for a year Created 10,000 images. HOW TO USE EDITOR: Go to the home page Select “Edit” tab Upload an img or paste URL Select or expand editable areas Write a text prompt Generate QUICK THOUGHTS EDITOR: It’s nice to edit within MJ MJ Gen quality > Photoshop Gen Fill But the UX/UI is clunky  Text prompting isn’t particularly coherent The editor doesn’t work well on small objects Unless you’re being very crafty Using REF Images (IMG/SREF/CREF) Text Prompts To help guide the output of selected areas Images of “me” keep getting flagged So I can’t edit a majority of them The content moderation is tight (as it should be) This is essentially a beta release I expect them to iron out the wrinkes Will this “kill” photoshop…no  (Too many of those posts) They only opened it up to the heavy users  Because it needs work. Again, it’s a few days old more creative uses will come. Running some more tests to find more uses. PS: Will do a separate post on the NEW Retexture feather. #midjourney #minimax #runwayml #ai #innovation

Runway ACT-1 → Finance Bro Babies. Took it for a test drive. And tried recreating the old  E*Trade talking baby commercials. It’s not perfect…but  The lip-sync/motion capture improvement Is definitely noticeable. Here’s how to use it ↓ PROCESS: 01. Open Runway Act-1 02. Record or upload a ref video 03. Create or upload a ref image 04. Generate (no prompt needed) MIDJOURNEY PROMPT: photo of a baby sitting in walker, fuzzy hair, chubby cheeks, wearing a horizontally striped onesie, looking directly at the camera, in a typical suburban household, shot on iPhone 13 --ar 5:3 --p --stylize 200 --v 6.1 QUICK THOUGHTS: The subtle details are improved Eye/mouth movement is tracked well Even the pupil/eyelid movement Lips still have some pixelation There’s subtle morphing in bottom/left corner The gens get softer after rendering (Like the hair/skin/lips) Render time is 3-4 mins (not bad) I exaggerated my motions for more effect I suck at acting More tests to come. Have fun. PS: Join the party → Subscribe to our newsletter for more weird stuff lol…LINK IN COMMENTS. #runwayml #minimax #midjourney #ai #innovation

Minimax is drunk → And it’s awesome. So stupid...but I can't stop. >>This is purely for entertainment<< To me…this is one way to learn  The IMG-2-VID tools Like Minimax, Runway, Kling, etc. Take one still image… And bring it to life in various ways. TRY MULTIPLE: Camera motions Shot types Character movements/actions Outcomes Here’s all the prompts used ↓ MJ PROMPT: movie still of a comedy film that was never made, and movie we never needed lol --chaos 10 --p --stylize 200 MINIMAX STRUCTURE: [camera motion] as [characters action/motion] [outcome] MINIMAX PROMPT EXAMPLES: 01. 2 men walk towards a bodega counter, offering the cashier an ice cream cone. 02. The camera pans down as a peacock walks through a bodega by 2 men 03. The camera pans right as 2 men walk right towards the bodega counter to offer the cashier a kitten 04. The camera pans down to reveal 2 men wearing clown shoes as they start to dance 05. The camera follows as 2 men shake pom-poms 06. Static shot as 2 men turn to eachother and point at each other awkwardly 07. Static shot as 2 men slowly raise pumpkins into the frame awkwardly 08. The camera pans right as 2 men walk right before shaking hands with a goofy alien 09. The camera whip-pans right to two children in goofy suits wearing oversized sunglasses with their arms crossed 10. The camera dolly-pushes in on the man's awkward terrified face lol QUICK THOUGHTS: Minimax keeps the image style Even when creating new scenery It feels like it has comedic timing (I know it doesn’t…but seriously) It’s easier to learn the tools While they are developing  Before there’s too many features Have fun. PS: Come nerd out with us on the MIDJOURNEY FAST HOURS PODCAST → LINK IN COMMENTS #minimax #runwayml #midjourney #ai #innovation

Minimax → Unintentionally Funny. The outtakes are better than the intended. (Borderline absurd) We’ll keep it light on a Saturday. **This is for entertainment purposes only** MJ PROMPT STRUCTURES: 01. movie still of a comedy film that was never made, and a movie we never needed lol --chaos 20 --ar 5:3 --p --v 6.1 02. found footage of an unknown [creature], on the deck of a Chinese fishing boat, surrounded by men proud of their catch --ar 4:5 --stylize 75 --v 6.1 03. Vintage {Trail cam, camcorder} footage of [creature/characters], --p --ar 5:3 --v 6.1 MINIMAX STRUCTURE: [camera motion] as [characters/subject] [action/motion] MINIMAX PROMPT EXAMPLES: 01. the camera dolly pushes in on bigfoot looking around then waving awkwardly at the camera 02. shaky hand-held footage of a sea creature flopping on the deck of a ship comedically causing a man to jump into the ocean 03. Static shot of a funny man blowing up a balloon animal 04. Tracking shot of 2 men walking towards a bodega counter, offering the cashier ice cream QUICK THOUGHTS: It feels like Minimax has a sense of humor (I know it doesn’t…but seriously) While jacked…the animation will improve Bigfoot footage is actually “awkward” (as intended) It’s easier to learn the tools If you’re doing “fun” things No agenda…no projects…no necessity It doesn’t feel like people share their fails. And that’s the reality of these tools. It’s never a one-click solution. PS: We’re nerding out and having some fun with it on Midjourney : Fast Hours...LINK IN COMMENTS. #minimax #runwayml #midjourney #ai #innovation

Runway Keyframes → Transformations. 01. Child → Adult 02. Chimp → Neanderthal 03. Heavy → Toned 04. Vintage → Modern Far from perfect. But ideas to play with. Used a combo of Remix Edit in MJ for images. PROCESS: 01. Gen base images in MJ 02. Remix/Edit in MJ 03. Animate in Runway MJ PROMPT EX: editorial photo, epic side profile shot of a child riding a bike through the streets of Rome, contrasting colors, spinning wheels, deep textures --ar 5:3 --p --stylize 200 --v 6.1 MJ REMIX PROMPT EX: editorial photo, epic side profile shot of a man riding a insane motorcycle through the streets of Rome, spinning wheels, contrasting colors, deep textures --ar 5:3 --p --stylize 200 --v 6.1 RUNWAY PROMPT EX: child riding a bicycle to the right as he seamlessly transforms into a grown man riding a motorcycle, 35mm style QUICK THOUGHTS: There’s inconsistencies  The base images set the stage. If they’re not visually consistent. The videos will warp or not move. To enhance motion use EDIT in MJ  To change the perspective/composition Nascar Img1: Positioning the car in bottom right Nascar Img: Positioning the car in the bottom left  It gives runway some room to fill in the blank Have Fun. PS: We’re nerding out on stuff like this here → Midjourney : Fast Hours. LINK IN COMMENTS #runwayml #midjourney #ai #chatgpt #innovation

Comically bad → But telling. Minimax is different. Tested some more complex ideas And the results were interesting. Farrrr from perfect or useable But the improv/speed is noticeable. And this tool is less than 2 months old. Most vid tools have issues with fast action Like fight scenes or dynamic camera/object motion But this is will probably improve quickly. PROBLEM AREAS: Extreme warping Altered physics  Unrealistic motion Pixelation/artifacts THINGS TO NOTICE: Multiple camera movements (in 1 shot) Fluid shot motion Dynamic body movements  Extremely fast motion/pacing The scenes with water MIDJOURNEY PROMPT: DVD still from 1980s dark fantasy adventure ninja movie, fight scene --ar 5:3 --stylize 200 --v 6.1 MINI-MAX PROMPT: epic violent fight scene, the camera follows as two ninjas punch and kick, fast violent action as the two battle to the death, cinematic brilliance QUICK THOUGHTS: You can look at this 2 ways “Wow” or “this is trash” It’s is very easy to dismiss “This is trash” mentality is understandable It looks like a bad meme But Minimax is a raw tool  And it has some impressive pieces Most (if not all) tools cannot handle this. We’ll see where it goes. #minimax #runwayml #ai #midjourney #innovation 

Runway Keyframes Test → Transformations. With some help from Midjourney Photoshop. (Not perfect) But a concept to play with. Here’s the process for the “age-transition” shot. PROCESS: 01. Gen base images in MJ. 02. Use Remix/edit for transformation. 03. Fine-tune in Photoshop. 04. Generate in Runway MJ PROMPT EX: editorial photo, medium shot, front view of a 20 year old man walking towards us in old Cappadocia, dirty blonde hair and beard, warm fall colors, epic architecture, high contrast, high precision photography --ar 5:3 --p --stylize 50 --v 6.1 MJ REMIX PROMPT: editorial photo, medium shot, front view of a 50 year old man walking towards us in old Cappadocia, dirty blonde hair and beard, warm fall colors, epic architecture, high contrast, high precision photography --ar 5:3 --p --stylize 50 --v 6.1 PHOTOSHOP EDITS: 01. Highlight elderly man and cut 02. Scale down the background 03. Gen Fill to add background padding 04. Paste old man back in 05. Erase/Feather/Gen Fill to integrate RUNWAY KEYFRAMES: First Keyframe: Young man Last Keyframe: Old Man RUNWAY PROMPT: the scene starts with a young man walking towards the camera as he seamlessly transforms into an older man PROBLEM AREAS: Subtle background morphing Inconsistent walking pace Movement feels stiff Static facial expressions QUICK THOUGHTS: MJ: The base images are paramount MJ: Use Remix in MJ for subtle variance It will keep the image structure RW: Too much variance = weird gen RW: Too little variance = no motion  RW: Text prompting directs the scene RW: Focusing more on the character object motion RW: Gives you more influence on the output Have fun. PS: Going deeper on some of this stuff on the Midjourney : Fast Hours pod…LINK IN COMMENTS #runwayml #midjourney #chatgpt #ai #innovation

A new week means a *new* Midjourney Fast Hours podcast. On round two of our Midjourney After Dark, Rory & Drew get into... ✔ AI Video Wars ✔ Future Possibilities ✔ Old Prompt Revival ✔ Community Curation ✔ The Creative Process ✔ Style Code Exploration & More Where to watch & listen: Apple: https://t.ly/toFps Spotify: https://t.ly/UAB9w YouTube: https://t.ly/Ed0gx #midjourney #aiimagegeneration #midjourneypodcast

Minimax → Very Interesting. Finally got around to testing it… It’s solid. It’s a bit raw. It has its pro’s and con’s. WHERE IT’S GOOD: The motion is mostly natural and fluid. More complex motion is possible. It has the “fastest-paced” motion  Of all the vid generators IMO. It’s improving quickly. WHERE IT’S BAD: A lot of glitching/morphing with faster motion Faces are not particularly consistent At further distances (This could also be a source image problem) Typically long generation times (3-4 mins) Prompt coherence is good But not super consistent Somewhat of a slot machine in terms of gens. 6-second clips at 720p res. MINIMAX PROMPTS USED: 01. zooming in shot of a man eating arguala from a bowl 02. pulling in shot as a donkey walks towards a young girl and nuzzles her face, subtle breeze blowing hair, natural dynamic motion 03. cinematic shot of a man walking up a hill to the right as the camera sharply focuses on him, farm animals grazing in the background 04. The camera motions to the right, man walking away from us towards an old house as the camera focuses sharply on him, subtle breeze blows the high grass 05. Emotional shot of a cowboy embracing his cow lovingly PROBLEM AREAS: CLIP 1: Trippy background plant motion CLIP 2: Not much CLIP 3: Glitchy steps, morphing objects CLIP 4: Hat/Face morph CLIP 5: Patting hand morph QUICK THOUGHTS: Far from perfect & hard to predict Some gens are solid first go Some gens take multiple re-rolls  Ref images text prompts = important Text prompts influence camera object motion The “Discover” page has some good insights Was most impressed with the donkey cow shots They have a free trial $9.99/month to start is reasonable Will be testing this more. Have Fun. PS: Quick tip…If you have a subscription to KREA…you can test all the video models (Minimax/Runway/Luma/Kling) #minimax #runwayml #midjourney #ai #chatgpt #innovation