FollowFox blog

Dreambooth Face Test 3 - Source Checkpoints: 1.4 vs 1.4-full-ema vs 1.5 vs 1.5-emaonly

Learn AI

About FollowFox is an AI exploratory initiative of the boutique marketing agency
Until AI takes over, offers a full range of marketing services at boutique quality by top talent in the region. Support us by:
  • Checking out our website link
  • Liking our LinkedIn page link
  • Checking our sortlist profile link
  • or by subscribing to this blog

Spoiler: Summary of findings

We think all base models did a fairly good job. If you are looking for a single recommendation: use SD 1.5-pruned at 3,000 steps. If you want to experiment a bit with steps or other settings, SD 1.5-pruned-emaonly seems like the best option. And in general, 1.5 weights did better than the 1.4 ones.

Overview and Setup

One of the first decisions you must make before doing a Dreambooth fine-tune is which model to use as a base. We decided to test a few official ones and compared the results to each other. The main goal was to find the model with the best output quality, but some performance optimization possibilities were also monitored (file size, training time, etc.)

Setup Details

Overall, the setup was identical to the previous experiment. See the details link. The main difference is that we tried four different source checkpoints at three different learning steps each (2k, 3k, 4k). Checkpoints tested:
  • sd-v1-4.ckpt link
  • sd-v1-4-full-ema.ckpt link
  • v1-5-pruned-emaonly.ckpt link
  • v1-5-pruned.ckpt link

Judging Criteria

As we are trying to standardize judging criteria, we used the same three concepts for each model. See details here the link:
  1. A realistic, high-quality photo
  2. Stylized avatar
  3. Fine-tuned subject depicted as Superman

Summary of Results

Overall, we got a lot of good fine-tuned checkpoints. The realistic photo was an easy task for all the models. With avatars, all did fine except for SD 1.4 Full ema. And finally, for Superman… we are not getting consistently good results, but if any, SD 1.5-pruned-emaonly did the best.

Loss Graphs and Training Time

Interestingly, both 1.5 checkpoints took a bit less time (46 mins to 4k steps) than 1.4 ones (51mins to 4k steps)
As for loss graphs, surprisingly, we got 4 identical graphs about loss and steps values.

Detailed Results (ranked from best to worst)

1st place: SD 1.5-pruned, 3,000 steps

2nd place: SD 1.5-pruned-emaonly, 4,000 steps

3rd place: SD 1.5-pruned-emaonly, 2,000 steps

4th place: SD 1.5-pruned-emaonly, 3,000 steps

5th place: SD 1.5-pruned, 4,000 steps

6th place: SD 1.4, 4,000 steps

7th place: SD 1.4, 2,000 steps

8th place: SD 1.4, 3,000 steps

9th place: SD 1.4 Full ema, 4,000 steps

10th place: SD 1.5-pruned, 2,000 steps

11th place: SD 1.4 Full ema, 2,000 steps

12th place: SD 1.4 Full ema, 3,000 steps

Made on