Crossbreeding Guide
How the AI breeding system works, why results vary so much, and the strategies experienced breeders use to produce elite hybrids.
How breeding works
When you breed two creatures, Wild Hybrids sends both animals to Google Gemini, our AI backbone. Gemini analyses the visual and biological characteristics of each parent and generates a new creature that plausibly combines them.
The AI produces a name, description, species label, rarity, and a unique image for the hybrid. No two breed runs produce the same result — even breeding the same pair twice will give you a different creature.
Step-by-step
- Go to Breed and select your first animal (Parent A). You can pre-select it from the collection detail page with the ↺ Breed button.
- Select a second animal (Parent B) from your collection using the picker.
- Click Generate. The system fires 4 parallel AI requests — one for each variant — so all four results arrive at roughly the same time.
- Review the 4 variants. Click any variant to expand it. When you've found one you like, click Keep.
- Kept creatures cost 1 token and appear in your collection immediately. Discarded variants are free.
The 4 variants
Each breed run produces 4 independent variant images of the same conceptual hybrid. They share the same parents and the same AI-generated name/description, but the visual representation differs between variants.
- Variants can range from incredibly detailed and photorealistic to cartoonish or abstract.
- One variant may focus heavily on Parent A while another leans toward Parent B.
- Occasionally all 4 are disappointing — that is the nature of generative AI. Discard and try again.
Tokens & cost
- Generating the 4 variants is free.
- Keeping a variant costs 1 token.
- Discarding all variants costs nothing.
- You receive a free daily token allowance. Check the Profile page for your remaining balance.
Rarity & offspring
The rarity assigned to a hybrid is determined by the AI based on a combination of parent rarities and the resulting creature's perceived power. The exact rarity is not guaranteed — it's probabilistic.
General tendencies
- Common × Common → Usually Common, occasionally Rare.
- Rare × Rare → Typically Rare, sometimes Epic.
- Epic × Epic → Often Epic, occasionally Legendary.
- Legendary × Legendary → Frequently Legendary or Epic.
- Mixed rarities → Result usually falls between or at the higher parent's tier.
Multi-generation breeding
You can breed a hybrid with another hybrid (or a library animal). There is no limit to the number of generations.
Why multi-gen?
- Combining hybrids produces the most unusual and visually striking results — features from 3 or 4 source animals blend together.
- A Legendary hybrid bred from two Epics can then be bred with another Legendary to try for an exceptional offspring.
- The Lineage panel on each creature's detail page shows its full family tree up to 2 generations back.
Combination tips
Not all pairings are created equal. Here are combinations that tend to produce the most interesting results:
What tends to produce poor results
- Two very similar animals (e.g. two different dog breeds) — the result is often indistinguishable from either parent.
- Very abstract or small animals paired with large complex ones — the smaller animal's features often get lost.
Quality issues & what to expect
AI image generation is imperfect. Here's an honest overview of what you'll encounter:
- ~70% of runs produce results that are uninspiring, anatomically odd, or don't clearly represent both parents.
- ~20% of runs produce decent hybrids that are worth keeping for collection or battle purposes.
- ~10% of runs produce genuinely striking creatures — these are the ones worth sharing to the feed.
Common artefacts to watch for
- Extra limbs — the AI sometimes generates anatomically impossible creatures with 6+ limbs.
- Blurry or smeared textures — especially in multi-gen hybrids with complex lineages.
- Wrong background — the image may show a background when you'd prefer a plain one (or vice versa).
- Disconnected features — one parent's features may appear floating or pasted rather than integrated.
None of these are errors in the system — they are inherent to the current state of AI image generation. When you see them, simply discard and regenerate.