[EveryDream trainer](https://github.com/victorchall/EveryDream-trainer) no longer requires cropped images. You only need to crop to exclude stuff you don't want trained, or to improve the portion of face close ups in your data. The EveryDream trainer now accepts multiple aspect ratios and can train on them natively.
But if you do wish to crop for other trainers, you can use [Birme](https://www.birme.net/?target_width=512&target_height=512&auto_focal=false&image_format=webp&quality_jpeg=95&quality_webp=99) to crop and resize first. There are various tools out there for this.
The default behavior will simply name the file the caption .EXT and, if needed, add _n at the end to avoid collisions, for use with EveryDream trainer or Kane Wallmann's dream booth fork.
"mrwho" or "joepenna" will add \[number\]@ as a prefix for use with MrWho's captioning system (on JoePenna dream both fork) which uses that naming standard to avoid file name collisions.
"txt" or "caption" will create a ".txt" or ".caption" file instead of renaming the image. ".txt" sidecar is another option for EveryDream trainer instead of getting the caption from the filename itself, and ".caption" is an option for other trainers.
You may find the following setting useful to deal with issues with bad auto-captioning. Start with defaults, and if you have issues with captions that seem inaccurate or reptitious try some of the following settings.
Uses an alternative "nucleus" algorithm instead of the default "beam 16" algorithm. Nucleus produces relatively short captions but reliably absent of repeated words and phrases, comparable to using beam 16 which can be adjusted further but may need more tweaking.
For the default beam 16 algorithm it limits the ability of words and phrases to be repeated. Higher value reduces repeated words and phrases. 0.6-1.4 are sensible values for beam 16. Default is 1.0 and works well with the defaulted value min_length of 24. Consider using higher values if you use a min_length higher than 24 with beam 16.
For nucleus (--nucleus), it simply changes the opinion on the prompt and does not impact repeats. Values ranging from 0.01 to 200 seem sensible and default of 1.0 usually works well.
![Beam vs Nucleus](../demo/beam_vs_nucleus_2.webp)
### --min_length
Adjusts the minimum length of prompt, measured in tokens. **Only applies to beam 16.** Useful to adjust along with --q_factor to keep it from repeating.
Default is 22. Sensible values are 15 to 30, max is 48. Larger values are much more prone to repeating phrases and should be accompanied by increasing --q_factor to avoid repeats.
If you continue to both increase min_length and q_factor with default beam algorithm in an attempt to get a really long caption without repeats it will generate oddly specific prompts. For example using the above image:
*"a painting of a group of people sitting at a table in a room with red drapes on the walls and gold trimmings on the ceiling, while one person is holding a wine glass in front of the other hand"*