lodestones/Chroma · Discussions

#116 opened 6 months ago by

JonathanZ

Just wanted to say THANK YOU!

❤️ 4

#115 opened 6 months ago by

olivetty

Model difference

👍 5

#113 opened 7 months ago by

Hikarias

Broken output on AMD 7900 XT

11

#112 opened 8 months ago by

CurieKuri

Color bands/frames on images & orange peel texture on skin

#111 opened 8 months ago by

AKDesign

Finetune Florence 2 model for Chroma.

👍 4

#110 opened 8 months ago by

Jester2025

Is there a new workflow? Including high-definition image upgrades, loading multiple Loras, and repairing faces, hands, and feet? Thank you.

➕🚀 4

#109 opened 8 months ago by

wzgrx

How do you achieve a realistic result?

4

#108 opened 8 months ago by

Prompts make little difference?

👍 2

5

#107 opened 8 months ago by

countsponge01

DMD2

🔥 1

#106 opened 8 months ago by

Ameenroayan

Quality in Base vs with Hyper lora

#105 opened 8 months ago by

citrusboi

Which version for Lora training?

👍 3

4

#104 opened 8 months ago by

AlbertoO

Realistic Character Lora training for Chroma?

#103 opened 9 months ago by

Nipaa

Author tags and styles.

#102 opened 9 months ago by

notafraud

Comparisons between v48, v48-detail-calibrated, v49, Chromad-HD, and Chroma-HD-Annealed

👍❤️ 10

9

#101 opened 9 months ago by

Struggling with v50

😔 3

11

#100 opened 9 months ago by

jire94ur

Final version?

18

#99 opened 9 months ago by

Fgdfgfthgr

Where V49

#98 opened 9 months ago by

Style reference feature

🧠👍 5

#97 opened 9 months ago by

val1106

Nunchaku Support possible?

🚀 1

#96 opened 9 months ago by

Euchale

What is the next high priority support?

#83 opened 9 months ago by

thesby

Why does my image look like a mess at CFG: 5?

#82 opened 9 months ago by

Is the model degrading in censorship?

🤯👀 4

#81 opened 9 months ago by

FRW43

Per-subject tags grouping.

7

#79 opened 9 months ago by

notafraud

How does the FLASH version work?

7

#78 opened 9 months ago by

NicoNick

Aesthetic 1girl meets Chroma (LoRa)

❤️👀 8

5

#77 opened 9 months ago by

Ainonake

T5 (v1.1) vs FLAN?

#76 opened 9 months ago by

roe2

Is v50 still the final objective?

❤️ 1

#75 opened 9 months ago by

scheduler, sampler, lora strength, and CFG settings for flux-dev loras on Chroma

#74 opened 9 months ago by

mikebig90

I added support for LoRA in Chroma trained with ai-toolkit in mlx-chroma.

🧠🔥 4

#73 opened 9 months ago by

jack813liu

"aesthetic 11" tag

👍🔥 1

4

#72 opened 9 months ago by

Kaleidia

Adding "seed numbers" for reproducible results

👀 1

#71 opened 10 months ago by

JKM2025

It`s a good model,a small suggestion

#70 opened 10 months ago by

zczcgery

Integrate builds in diffusers format in pipeline?

👀 1

#69 opened 10 months ago by

tintwotin

Prompting

#68 opened 10 months ago by

yamatazen

`few-steps-rl` is the right direction

#65 opened 10 months ago by

notafraud

Replace t5xxl with other LLM ?

👀 3

6

#64 opened 10 months ago by

thesby

Speed of generation Compare to 4 step flux shnell ?

#63 opened 10 months ago by

abhinandan1111

请问最终这个模型会训练到多少轮？现在40了，或者预估训练到什么时候？我正在本地部署Lora训练。谢谢

#62 opened 10 months ago by

yiwualang

Is there a way to achieve the effect of flux kontext

#61 opened 10 months ago by

michaelj

Feet issues

10

#60 opened 10 months ago by

Hellrunner

Attention_mask issue happening when i try to run chroma

#59 opened 10 months ago by

leonvizard

I implemented Chroma's text_to_image inference using Apple's MLX

❤️ 8

#56 opened 10 months ago by

jack813liu

Small Suggestion

👍 2

#54 opened 11 months ago by

loremipsum2256

LoRA Training

👍 4

#53 opened 11 months ago by

dthz7iz7

vram minimal requirement