Datasets:

TheMindExpansionNetwork
/

psychadeli-morphing-menagerie

Error code:   StreamingRowsError
Exception:    ValueError
Message:      Expected object or value
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
                  pa_table = paj.read_json(
                      io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
                  )
                File "pyarrow/_json.pyx", line 342, in pyarrow._json.read_json
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: JSON parse error: Column(/115/inputs/megapixels/[]) changed from string to number in row 0
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 506, in __iter__
                  yield from self.ex_iterable
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 398, in __iter__
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 304, in _generate_tables
                  batch = json_encode_fields_in_json_lines(original_batch, json_field_paths)
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 111, in json_encode_fields_in_json_lines
                  examples = [ujson_loads(line) for line in original_batch.splitlines()]
                              ~~~~~~~~~~~^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
                  return pd.io.json.ujson_loads(*args, **kwargs)
                         ~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
              ValueError: Expected object or value

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

PSYCHADELI — The Morphing Menagerie

FICTIONAL / AI-GENERATED. Eleven original funny creature concepts, eleven unmodified generated videos, and eleven locally finished silent visual loop derivatives. No real animals were filmed. No live rendering service is offered.

Visit the gallery · Creative captions · Exact manifest

Download

hf download TheMindExpansionNetwork/psychadeli-morphing-menagerie --repo-type dataset --local-dir psychadeli

Inventory

originals/: exactly 11 byte-preserved output MP4 files from the supplied ComfyUI workflow. Generated audio, where present, is unchanged.
loops/: exactly 11 silent H.264/yuv420p MP4 cycle edits, each 360 frames, 24 fps, exactly 15.000 seconds.
thumbnails/: 11 actual frames extracted from the silent clips.
seams/: boundary contact sheets and per-clip numerical continuity measurements.
workflows/: 11 sanitized API graphs. No endpoint URLs, tokens, private paths or model weights.
captions.json / captions.md: original fictional character stories, intended morphs, exact prompts and unique seeds. Creative intentions are not guaranteed literal frame annotations.
manifest.json: prompt IDs, hashes, actual source/video/audio properties and processing measurements.
SHA256SUMS: checksum of every other released file.

Provenance and source contract

Generation uses the user's supplied VDN-H3-VS-fastvideoH3_t2Swing_USinners_API.json graph, kept immutable in the production workspace. Its SHA-256 is 460957d0fe3cc444e851835aebcda4fbd484c80049713b6384e1b6b1547fb02c.

The only per-job modifications were input prompt node 6005, duration node 221 (15 seconds), random seed node 6141 and filename prefix node 6170. Every exact prompt begins with literal swingusinnerz. Source target: 16:9, 0.8 megapixels, dimensions selected in multiples of 32, 24 fps. The H3 temporal grid is 17k+5: 15 requested seconds becomes 362 frames, approximately 15.083333 seconds of video. Container or audio durations can differ slightly; measured streams are in the manifest.

Preserved model stack:

minimax_h3_fl2va_pruned_int8_convrot.safetensors
qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
minimax_h3_video_vae_fp16.safetensors
minimax_h3_audio_vae_fp32.safetensors
swingusinnerz_t2v_minimax_h3_v1.safetensors, strength 1
ApplyVDNH3Advanced, checkpoint stage-dmd-step-250, turbo adapter enabled, both anchor frames, grouped attention, unchanged source optimizations
8 steps, Euler/simple, sigma shifts video 12 / audio 3, unchanged chunk/torch settings

No model weights are distributed. These are new original character concepts using a supplied third-party style adapter; the style trigger is not proof of ownership of an artistic tradition or training material.

Exact loop processing

Zero-based source indices: take body frames 36–325 (290 frames). Blend tail 326–361 with head 0–35 across 36 frames, with head weight increasing from 0 to 1. The default is linear weight t=(N-1)/35; clips identified in their manifest with a local refinement use quintic easing w=6t^5-15t^4+10t^3 to reduce visible residual silhouettes near overlap endpoints. Both methods can still show midpoint superposition. Append that blend to the body, producing a 326-frame cyclic sequence. The sequence ends at original frame 35 and restarts at frame 36, placing the repetition boundary at a naturally adjacent source-frame transition rather than at the uncorrected original end/start.

Expand presentation time by 360/326 and resample to 24 fps with FFmpeg fps, using frame duplication, not motion interpolation. Trim to 360 frames. Encode H.264 CRF18, yuv420p, faststart. Remove audio only from the derivative. No reversal / ping-pong, optical-flow interpolation, or additional generation was used. Source motion is approximately 9.4% slower after time expansion; cadence includes duplicate frames. Original videos remain unmodified.

This is deliberate cyclic overlap editing, not HTML repeat alone. It is also not a guarantee of perceptual seamlessness: crossfades can create ghosting, source lighting/composition can drift despite prompts, and duplicated frames can be perceptible. See contact sheets and visual review notes rather than treating a low boundary score as proof.

Seam measurement

Decode each final derivative to RGB at 192×108. Compute mean absolute RGB difference (0–255 scale) between the last and first decoded frames. Compare with median and 95th percentile of all internal adjacent-frame differences. The manifest records raw values and boundary/median ratio. Downsampling and static background can make scores look better than perceived foreground continuity. Boundary contact sheets show frames 354, 357, 359, 0, 1, 3, 12 and 36. Still review cannot establish temporal fluidity or audio quality.

Rights, attribution and restrictions

Read the official MiniMax H3 license before use. Its default terms contain territorial restrictions (EU, UK, Republic of Korea and USA are excluded territories), restrictions that can extend to outputs, commercial conditions including the USD20M revenue threshold, attribution requirements and restrictions on using outputs to improve other AI models. Separate written permissions may apply; this collection does not represent that any recipient has them. Public availability does not override those terms.

The supplied style adapter's training-data provenance and separate license were not independently established. We do not claim its training materials are original, licensed for every use or free of third-party rights. Original prompt/character direction does not clear model, adapter or incidental resemblance rights. No unrestricted commercial license is granted here. Review applicable model/adapter terms and obtain permissions where required. This is provenance disclosure, not legal advice.

Intended use and limitations

Artistic browsing, study of cyclic editing and reuse only where allowed by applicable rights and territorial terms. These are fictional animation outputs, not evidence of real events. Not suitable as scientific animal imagery or as a verified dataset of exact prompt compliance. There may be incidental cartoon vocalizations in originals; audio intelligibility and content have not been comprehensively reviewed. Silent derivatives contain no audio stream.

Residents

Ruby Rumba — A berry-headed busker pays rent in tiny tap dances.
Bellows Belle — The night orchestra has misplaced its conductor. Belle is all the instruments.
Earl Wobble — A deep-sea tea critic finds every ocean slightly under-steeped.
Spore Snort — This truffle detective keeps accidentally becoming the evidence.
Nimbus Noodle — The weather forecast is a small chance of socks.
Professor Peek — An astronomer discovering that the universe is mostly its own elbow.
Peel Puff — Guardian of one extremely unimportant fruit bowl.
Iris Oops — A garden gossip who has literally seen it all.
Orbit Ribbit — The smallest planetarium with the loudest opinions.
Silk Swerve — A swimming gift wrap that refuses to be anybody’s present.
Mallow Major — Mission control has requested more cocoa and less orbital frosting.

Downloads last month: 3

Total file size:

130 MB