-
Notifications
You must be signed in to change notification settings - Fork 345
/
_vision_models.py
1606 lines (1353 loc) · 59.5 KB
/
_vision_models.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Copyright 2023 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# pylint: disable=bad-continuation, line-too-long, protected-access
"""Classes for working with vision models."""
import base64
import dataclasses
import hashlib
import io
import json
import pathlib
import typing
from typing import Any, Dict, List, Literal, Optional, Union
import urllib
from google.cloud import storage
from google.cloud.aiplatform import initializer as aiplatform_initializer
from vertexai._model_garden import _model_garden_models
# pylint: disable=g-import-not-at-top
try:
from IPython import display as IPython_display
except ImportError:
IPython_display = None
try:
from PIL import Image as PIL_Image
except ImportError:
PIL_Image = None
_SUPPORTED_UPSCALING_SIZES = [2048, 4096]
class Image:
"""Image."""
__module__ = "vertexai.vision_models"
_loaded_bytes: Optional[bytes] = None
_loaded_image: Optional["PIL_Image.Image"] = None
_gcs_uri: Optional[str] = None
def __init__(
self,
image_bytes: Optional[bytes] = None,
gcs_uri: Optional[str] = None,
):
"""Creates an `Image` object.
Args:
image_bytes: Image file bytes. Image can be in PNG or JPEG format.
gcs_uri: Image URI in Google Cloud Storage.
"""
if bool(image_bytes) == bool(gcs_uri):
raise ValueError("Either image_bytes or gcs_uri must be provided.")
self._image_bytes = image_bytes
self._gcs_uri = gcs_uri
@staticmethod
def load_from_file(location: str) -> "Image":
"""Loads image from local file or Google Cloud Storage.
Args:
location: Local path or Google Cloud Storage uri from where to load
the image.
Returns:
Loaded image as an `Image` object.
"""
parsed_url = urllib.parse.urlparse(location)
if (
parsed_url.scheme == "https"
and parsed_url.netloc == "storage.googleapis.com"
):
parsed_url = parsed_url._replace(
scheme="gs", netloc="", path=f"/{urllib.parse.unquote(parsed_url.path)}"
)
location = urllib.parse.urlunparse(parsed_url)
if parsed_url.scheme == "gs":
return Image(gcs_uri=location)
# Load image from local path
image_bytes = pathlib.Path(location).read_bytes()
image = Image(image_bytes=image_bytes)
return image
@property
def _blob(self) -> storage.Blob:
if self._gcs_uri is None:
raise AttributeError("_blob is only supported when gcs_uri is set.")
storage_client = storage.Client(
credentials=aiplatform_initializer.global_config.credentials
)
blob = storage.Blob.from_string(uri=self._gcs_uri, client=storage_client)
# Needed to populate `blob.content_type`
blob.reload()
return blob
@property
def _image_bytes(self) -> bytes:
if self._loaded_bytes is None:
self._loaded_bytes = self._blob.download_as_bytes()
return self._loaded_bytes
@_image_bytes.setter
def _image_bytes(self, value: bytes):
self._loaded_bytes = value
@property
def _pil_image(self) -> "PIL_Image.Image":
if self._loaded_image is None:
if not PIL_Image:
raise RuntimeError(
"The PIL module is not available. Please install the Pillow package."
)
self._loaded_image = PIL_Image.open(io.BytesIO(self._image_bytes))
return self._loaded_image
@property
def _size(self):
return self._pil_image.size
@property
def _mime_type(self) -> str:
"""Returns the MIME type of the image."""
if self._gcs_uri:
return self._blob.content_type
if PIL_Image:
return PIL_Image.MIME.get(self._pil_image.format, "image/jpeg")
# Fall back to jpeg
return "image/jpeg"
def show(self):
"""Shows the image.
This method only works when in a notebook environment.
"""
if PIL_Image and IPython_display:
IPython_display.display(self._pil_image)
def save(self, location: str):
"""Saves image to a file.
Args:
location: Local path where to save the image.
"""
pathlib.Path(location).write_bytes(self._image_bytes)
def _as_base64_string(self) -> str:
"""Encodes image using the base64 encoding.
Returns:
Base64 encoding of the image as a string.
"""
# ! b64encode returns `bytes` object, not `str`.
# We need to convert `bytes` to `str`, otherwise we get service error:
# "received initial metadata size exceeds limit"
return base64.b64encode(self._image_bytes).decode("ascii")
class Video:
"""Video."""
__module__ = "vertexai.vision_models"
_loaded_bytes: Optional[bytes] = None
_gcs_uri: Optional[str] = None
def __init__(
self,
video_bytes: Optional[bytes] = None,
gcs_uri: Optional[str] = None,
):
"""Creates a `Video` object.
Args:
video_bytes: Video file bytes. Video can be in AVI, FLV, MKV, MOV,
MP4, MPEG, MPG, WEBM, and WMV formats.
gcs_uri: Image URI in Google Cloud Storage.
"""
if bool(video_bytes) == bool(gcs_uri):
raise ValueError("Either video_bytes or gcs_uri must be provided.")
self._video_bytes = video_bytes
self._gcs_uri = gcs_uri
@staticmethod
def load_from_file(location: str) -> "Video":
"""Loads video from local file or Google Cloud Storage.
Args:
location: Local path or Google Cloud Storage uri from where to load
the video.
Returns:
Loaded video as an `Video` object.
"""
parsed_url = urllib.parse.urlparse(location)
if (
parsed_url.scheme == "https"
and parsed_url.netloc == "storage.googleapis.com"
):
parsed_url = parsed_url._replace(
scheme="gs", netloc="", path=f"/{urllib.parse.unquote(parsed_url.path)}"
)
location = urllib.parse.urlunparse(parsed_url)
if parsed_url.scheme == "gs":
return Video(gcs_uri=location)
# Load video from local path
video_bytes = pathlib.Path(location).read_bytes()
video = Video(video_bytes=video_bytes)
return video
@property
def _blob(self) -> storage.Blob:
if self._gcs_uri is None:
raise AttributeError("_blob is only supported when gcs_uri is set.")
storage_client = storage.Client(
credentials=aiplatform_initializer.global_config.credentials
)
blob = storage.Blob.from_string(uri=self._gcs_uri, client=storage_client)
# Needed to populate `blob.content_type`
blob.reload()
return blob
@property
def _video_bytes(self) -> bytes:
if self._loaded_bytes is None:
self._loaded_bytes = self._blob.download_as_bytes()
return self._loaded_bytes
@_video_bytes.setter
def _video_bytes(self, value: bytes):
self._loaded_bytes = value
@property
def _mime_type(self) -> str:
"""Returns the MIME type of the video."""
if self._gcs_uri:
return self._blob.content_type
# Fall back to mp4
return "video/mp4"
def save(self, location: str):
"""Saves video to a file.
Args:
location: Local path where to save the video.
"""
pathlib.Path(location).write_bytes(self._video_bytes)
def _as_base64_string(self) -> str:
"""Encodes video using the base64 encoding.
Returns:
Base64 encoding of the video as a string.
"""
# ! b64encode returns `bytes` object, not `str`.
# We need to convert `bytes` to `str`, otherwise we get service error:
# "received initial metadata size exceeds limit"
return base64.b64encode(self._video_bytes).decode("ascii")
class VideoSegmentConfig:
"""The specific video segments (in seconds) the embeddings are generated for."""
__module__ = "vertexai.vision_models"
start_offset_sec: int
end_offset_sec: int
interval_sec: int
def __init__(
self,
start_offset_sec: int = 0,
end_offset_sec: int = 120,
interval_sec: int = 16,
):
"""Creates a `VideoSegmentConfig` object.
Args:
start_offset_sec: Start time offset (in seconds) to generate embeddings for.
end_offset_sec: End time offset (in seconds) to generate embeddings for.
interval_sec: Interval to divide video for generated embeddings.
"""
self.start_offset_sec = start_offset_sec
self.end_offset_sec = end_offset_sec
self.interval_sec = interval_sec
class VideoEmbedding:
"""Embeddings generated from video with offset times."""
__module__ = "vertexai.vision_models"
start_offset_sec: int
end_offset_sec: int
embedding: List[float]
def __init__(
self, start_offset_sec: int, end_offset_sec: int, embedding: List[float]
):
"""Creates a `VideoEmbedding` object.
Args:
start_offset_sec: Start time offset (in seconds) of generated embeddings.
end_offset_sec: End time offset (in seconds) of generated embeddings.
embedding: Generated embedding for interval.
"""
self.start_offset_sec = start_offset_sec
self.end_offset_sec = end_offset_sec
self.embedding = embedding
class ImageGenerationModel(
_model_garden_models._ModelGardenModel # pylint: disable=protected-access
):
"""Generates images from text prompt.
Examples::
model = ImageGenerationModel.from_pretrained("imagegeneration@002")
response = model.generate_images(
prompt="Astronaut riding a horse",
# Optional:
number_of_images=1,
seed=0,
)
response[0].show()
response[0].save("image1.png")
"""
__module__ = "vertexai.preview.vision_models"
_INSTANCE_SCHEMA_URI = "gs://google-cloud-aiplatform/schema/predict/instance/vision_generative_model_1.0.0.yaml"
def _generate_images(
self,
prompt: str,
*,
negative_prompt: Optional[str] = None,
number_of_images: int = 1,
width: Optional[int] = None,
height: Optional[int] = None,
aspect_ratio: Optional[Literal["1:1", "9:16", "16:9", "4:3", "3:4"]] = None,
guidance_scale: Optional[float] = None,
seed: Optional[int] = None,
base_image: Optional["Image"] = None,
mask: Optional["Image"] = None,
edit_mode: Optional[
Literal[
"inpainting-insert",
"inpainting-remove",
"outpainting",
"product-image",
]
] = None,
mask_mode: Optional[Literal["background", "foreground", "semantic"]] = None,
segmentation_classes: Optional[List[str]] = None,
mask_dilation: Optional[float] = None,
product_position: Optional[Literal["fixed", "reposition"]] = None,
output_mime_type: Optional[Literal["image/png", "image/jpeg"]] = None,
compression_quality: Optional[float] = None,
language: Optional[str] = None,
output_gcs_uri: Optional[str] = None,
add_watermark: Optional[bool] = None,
safety_filter_level: Optional[
Literal[
"block_most",
"block_some",
"block_few",
"block_fewest",
"block_low_and_above",
"block_medium_and_above",
"block_only_high",
"block_none",
]
] = None,
person_generation: Optional[
Literal["dont_allow", "allow_adult", "allow_all"]
] = None,
) -> "ImageGenerationResponse":
"""Generates images from text prompt.
Args:
prompt: Text prompt for the image.
negative_prompt: A description of what you want to omit in the generated
images.
number_of_images: Number of images to generate. Range: 1..8.
width: Width of the image. One of the sizes must be 256 or 1024.
height: Height of the image. One of the sizes must be 256 or 1024.
aspect_ratio: Aspect ratio for the image. Supported values are:
* 1:1 - Square image
* 9:16 - Portait image
* 16:9 - Landscape image
* 4:3 - Landscape, desktop ratio.
* 3:4 - Portrait, desktop ratio
guidance_scale: Controls the strength of the prompt. Suggested values
are - * 0-9 (low strength) * 10-20 (medium strength) * 21+ (high
strength)
seed: Image generation random seed.
base_image: Base image to use for the image generation.
mask: Mask for the base image.
edit_mode: Describes the editing mode for the request. Supported values
are -
* inpainting-insert: fills the mask area based on the text
prompt (requires mask and text)
* inpainting-remove: removes the object(s) in the mask area. (requires mask)
* outpainting: extend the image based on the mask area. (Requires
mask)
* product-image: Changes the background for the predominant
product or subject in the image
mask_mode: Solicits generation of the mask (v/s providing mask as an
input). Supported values are:
* background: Automatically generates a mask for all regions except
the primary subject(s) of the image
* foreground: Automatically generates a mask for the primary
subjects(s) of the image.
* semantic: Segment one or more of the segmentation classes using
class ID
segmentation_classes: List of class IDs for segmentation. Max of 5 IDs
mask_dilation: Defines the dilation percentage of the mask provided.
Float between 0 and 1. Defaults to 0.03
product_position: Defines whether the product should stay fixed or be
repositioned. Supported Values:
* fixed: Fixed position
* reposition: Can be moved (default)
output_mime_type: Which image format should the output be saved as.
Supported values:
* image/png: Save as a PNG image
* image/jpeg: Save as a JPEG image
compression_quality: Level of compression if the output mime type is
selected to be image/jpeg. Float between 0 to 100
language: Language of the text prompt for the image. Default: None.
Supported values are `"en"` for English, `"hi"` for Hindi, `"ja"` for
Japanese, `"ko"` for Korean, and `"auto"` for automatic language
detection.
output_gcs_uri: Google Cloud Storage uri to store the generated images.
add_watermark: Add a watermark to the generated image
safety_filter_level: Adds a filter level to Safety filtering. Supported
values are:
* block_most : Strongest filtering level, most strict
blocking
* block_some : Block some problematic prompts and responses
* block_few : Block fewer problematic prompts and responses
* block_fewest : Block very few problematic prompts and
responses
For Imagen 3.0 and Imagen 2.0 Editing (model_name:
`imagen-3.0-generate-001`, `imagen-3.0-fast-generate-001`,
`imagen-2.0-edit-preview-0627` and `imagegeneration@006`), the
following safety filter levels are supported:
* block_low_and_above : Block low and above safety scores
* block_medium_and_above : Block medium and above safety scores
* block_only_high : Block only high safety scores
* block_none : Block nothing
person_generation: Allow generation of people by the model Supported
values are:
* dont_allow : Block generation of people
* allow_adult : Generate adults, but not children
* allow_all : Generate adults and children
Returns:
An `ImageGenerationResponse` object.
"""
# Note: Only a single prompt is supported by the service.
instance = {"prompt": prompt}
shared_generation_parameters = {
"prompt": prompt,
# b/295946075 The service stopped supporting image sizes.
# "width": width,
# "height": height,
"number_of_images_in_batch": number_of_images,
}
if base_image:
if base_image._gcs_uri: # pylint: disable=protected-access
instance["image"] = {
"gcsUri": base_image._gcs_uri # pylint: disable=protected-access
}
shared_generation_parameters[
"base_image_uri"
] = base_image._gcs_uri # pylint: disable=protected-access
else:
instance["image"] = {
"bytesBase64Encoded": base_image._as_base64_string() # pylint: disable=protected-access
}
shared_generation_parameters["base_image_hash"] = hashlib.sha1(
base_image._image_bytes # pylint: disable=protected-access
).hexdigest()
if mask:
if mask._gcs_uri: # pylint: disable=protected-access
instance["mask"] = {
"image": {
"gcsUri": mask._gcs_uri # pylint: disable=protected-access
},
}
shared_generation_parameters[
"mask_uri"
] = mask._gcs_uri # pylint: disable=protected-access
else:
instance["mask"] = {
"image": {
"bytesBase64Encoded": mask._as_base64_string() # pylint: disable=protected-access
},
}
shared_generation_parameters["mask_hash"] = hashlib.sha1(
mask._image_bytes # pylint: disable=protected-access
).hexdigest()
parameters = {}
max_size = max(width or 0, height or 0) or None
if aspect_ratio is not None:
parameters["aspectRatio"] = aspect_ratio
elif max_size:
# Note: The size needs to be a string
parameters["sampleImageSize"] = str(max_size)
if height is not None and width is not None and height != width:
parameters["aspectRatio"] = f"{width}:{height}"
parameters["sampleCount"] = number_of_images
if negative_prompt:
parameters["negativePrompt"] = negative_prompt
shared_generation_parameters["negative_prompt"] = negative_prompt
if seed is not None:
# Note: String seed and numerical seed give different results
parameters["seed"] = seed
shared_generation_parameters["seed"] = seed
if guidance_scale is not None:
parameters["guidanceScale"] = guidance_scale
shared_generation_parameters["guidance_scale"] = guidance_scale
if language is not None:
parameters["language"] = language
shared_generation_parameters["language"] = language
if output_gcs_uri is not None:
parameters["storageUri"] = output_gcs_uri
shared_generation_parameters["storage_uri"] = output_gcs_uri
parameters["editConfig"] = {}
if edit_mode is not None:
parameters["editConfig"]["editMode"] = edit_mode
shared_generation_parameters["edit_mode"] = edit_mode
if mask is None and edit_mode != "product-image":
parameters["editConfig"]["maskMode"] = {}
if mask_mode is not None:
parameters["editConfig"]["maskMode"]["maskType"] = mask_mode
shared_generation_parameters["mask_mode"] = mask_mode
if segmentation_classes is not None:
parameters["editConfig"]["maskMode"]["classes"] = segmentation_classes
shared_generation_parameters["classes"] = segmentation_classes
if mask_dilation is not None:
parameters["editConfig"]["maskDilation"] = mask_dilation
shared_generation_parameters["mask_dilation"] = mask_dilation
if product_position is not None:
parameters["editConfig"]["productPosition"] = product_position
shared_generation_parameters["product_position"] = product_position
parameters["outputOptions"] = {}
if output_mime_type is not None:
parameters["outputOptions"]["mimeType"] = output_mime_type
shared_generation_parameters["mime_type"] = output_mime_type
if compression_quality is not None:
parameters["outputOptions"]["compressionQuality"] = compression_quality
shared_generation_parameters["compression_quality"] = compression_quality
if add_watermark is not None:
parameters["addWatermark"] = add_watermark
shared_generation_parameters["add_watermark"] = add_watermark
if safety_filter_level is not None:
parameters["safetySetting"] = safety_filter_level
shared_generation_parameters["safety_filter_level"] = safety_filter_level
if person_generation is not None:
parameters["personGeneration"] = person_generation
shared_generation_parameters["person_generation"] = person_generation
response = self._endpoint.predict(
instances=[instance],
parameters=parameters,
)
generated_images: List["GeneratedImage"] = []
for idx, prediction in enumerate(response.predictions):
generation_parameters = dict(shared_generation_parameters)
generation_parameters["index_of_image_in_batch"] = idx
encoded_bytes = prediction.get("bytesBase64Encoded")
generated_image = GeneratedImage(
image_bytes=base64.b64decode(encoded_bytes) if encoded_bytes else None,
generation_parameters=generation_parameters,
gcs_uri=prediction.get("gcsUri"),
)
generated_images.append(generated_image)
return ImageGenerationResponse(images=generated_images)
def generate_images(
self,
prompt: str,
*,
negative_prompt: Optional[str] = None,
number_of_images: int = 1,
aspect_ratio: Optional[Literal["1:1", "9:16", "16:9", "4:3", "3:4"]] = None,
guidance_scale: Optional[float] = None,
language: Optional[str] = None,
seed: Optional[int] = None,
output_gcs_uri: Optional[str] = None,
add_watermark: Optional[bool] = True,
safety_filter_level: Optional[
Literal["block_most", "block_some", "block_few", "block_fewest"]
] = None,
person_generation: Optional[
Literal["dont_allow", "allow_adult", "allow_all"]
] = None,
) -> "ImageGenerationResponse":
"""Generates images from text prompt.
Args:
prompt: Text prompt for the image.
negative_prompt: A description of what you want to omit in the generated
images.
number_of_images: Number of images to generate. Range: 1..8.
aspect_ratio: Changes the aspect ratio of the generated image Supported
values are:
* "1:1" : 1:1 aspect ratio
* "9:16" : 9:16 aspect ratio
* "16:9" : 16:9 aspect ratio
* "4:3" : 4:3 aspect ratio
* "3:4" : 3:4 aspect_ratio
guidance_scale: Controls the strength of the prompt. Suggested values are:
* 0-9 (low strength)
* 10-20 (medium strength)
* 21+ (high strength)
language: Language of the text prompt for the image. Default: None.
Supported values are `"en"` for English, `"hi"` for Hindi, `"ja"`
for Japanese, `"ko"` for Korean, and `"auto"` for automatic language
detection.
seed: Image generation random seed.
output_gcs_uri: Google Cloud Storage uri to store the generated images.
add_watermark: Add a watermark to the generated image
safety_filter_level: Adds a filter level to Safety filtering. Supported
values are:
* "block_most" : Strongest filtering level, most strict
blocking
* "block_some" : Block some problematic prompts and responses
* "block_few" : Block fewer problematic prompts and responses
* "block_fewest" : Block very few problematic prompts and responses
person_generation: Allow generation of people by the model Supported
values are:
* "dont_allow" : Block generation of people
* "allow_adult" : Generate adults, but not children
* "allow_all" : Generate adults and children
Returns:
An `ImageGenerationResponse` object.
"""
return self._generate_images(
prompt=prompt,
negative_prompt=negative_prompt,
number_of_images=number_of_images,
aspect_ratio=aspect_ratio,
guidance_scale=guidance_scale,
language=language,
seed=seed,
output_gcs_uri=output_gcs_uri,
add_watermark=add_watermark,
safety_filter_level=safety_filter_level,
person_generation=person_generation,
)
def edit_image(
self,
*,
prompt: str,
base_image: "Image",
mask: Optional["Image"] = None,
negative_prompt: Optional[str] = None,
number_of_images: int = 1,
guidance_scale: Optional[float] = None,
edit_mode: Optional[
Literal[
"inpainting-insert", "inpainting-remove", "outpainting", "product-image"
]
] = None,
mask_mode: Optional[Literal["background", "foreground", "semantic"]] = None,
segmentation_classes: Optional[List[str]] = None,
mask_dilation: Optional[float] = None,
product_position: Optional[Literal["fixed", "reposition"]] = None,
output_mime_type: Optional[Literal["image/png", "image/jpeg"]] = None,
compression_quality: Optional[float] = None,
language: Optional[str] = None,
seed: Optional[int] = None,
output_gcs_uri: Optional[str] = None,
safety_filter_level: Optional[
Literal["block_most", "block_some", "block_few", "block_fewest"]
] = None,
person_generation: Optional[
Literal["dont_allow", "allow_adult", "allow_all"]
] = None,
) -> "ImageGenerationResponse":
"""Edits an existing image based on text prompt.
Args:
prompt: Text prompt for the image.
base_image: Base image from which to generate the new image.
mask: Mask for the base image.
negative_prompt: A description of what you want to omit in
the generated images.
number_of_images: Number of images to generate. Range: 1..8.
guidance_scale: Controls the strength of the prompt.
Suggested values are:
* 0-9 (low strength)
* 10-20 (medium strength)
* 21+ (high strength)
edit_mode: Describes the editing mode for the request. Supported values are:
* inpainting-insert: fills the mask area based on the text prompt
(requires mask and text)
* inpainting-remove: removes the object(s) in the mask area.
(requires mask)
* outpainting: extend the image based on the mask area.
(Requires mask)
* product-image: Changes the background for the predominant product
or subject in the image
mask_mode: Solicits generation of the mask (v/s providing mask as an
input). Supported values are:
* background: Automatically generates a mask for all regions except
the primary subject(s) of the image
* foreground: Automatically generates a mask for the primary
subjects(s) of the image.
* semantic: Segment one or more of the segmentation classes using
class ID
segmentation_classes: List of class IDs for segmentation. Max of 5 IDs
mask_dilation: Defines the dilation percentage of the mask provided.
Float between 0 and 1. Defaults to 0.03
product_position: Defines whether the product should stay fixed or be
repositioned. Supported Values:
* fixed: Fixed position
* reposition: Can be moved (default)
output_mime_type: Which image format should the output be saved as.
Supported values:
* image/png: Save as a PNG image
* image/jpeg: Save as a JPEG image
compression_quality: Level of compression if the output mime type is
selected to be image/jpeg. Float between 0 to 100
language: Language of the text prompt for the image. Default: None.
Supported values are `"en"` for English, `"hi"` for Hindi,
`"ja"` for Japanese, `"ko"` for Korean, and `"auto"` for
automatic language detection.
seed: Image generation random seed.
output_gcs_uri: Google Cloud Storage uri to store the edited images.
safety_filter_level: Adds a filter level to Safety filtering. Supported
values are:
* "block_most" : Strongest filtering level, most strict
blocking
* "block_some" : Block some problematic prompts and responses
* "block_few" : Block fewer problematic prompts and responses
* "block_fewest" : Block very few problematic prompts and responses
person_generation: Allow generation of people by the model Supported
values are:
* "dont_allow" : Block generation of people
* "allow_adult" : Generate adults, but not children
* "allow_all" : Generate adults and children
Returns:
An `ImageGenerationResponse` object.
"""
return self._generate_images(
prompt=prompt,
negative_prompt=negative_prompt,
number_of_images=number_of_images,
guidance_scale=guidance_scale,
seed=seed,
base_image=base_image,
mask=mask,
edit_mode=edit_mode,
mask_mode=mask_mode,
segmentation_classes=segmentation_classes,
mask_dilation=mask_dilation,
product_position=product_position,
output_mime_type=output_mime_type,
compression_quality=compression_quality,
language=language,
output_gcs_uri=output_gcs_uri,
add_watermark=False, # Not supported for editing yet
safety_filter_level=safety_filter_level,
person_generation=person_generation,
)
def upscale_image(
self,
image: Union["Image", "GeneratedImage"],
new_size: Optional[int] = 2048,
upscale_factor: Optional[Literal["x2", "x4"]] = None,
output_mime_type: Optional[Literal["image/png", "image/jpeg"]] = "image/png",
output_compression_quality: Optional[int] = None,
output_gcs_uri: Optional[str] = None,
) -> "Image":
"""Upscales an image.
This supports upscaling images generated through the `generate_images()`
method, or upscaling a new image.
Examples::
# Upscale a generated image
model = ImageGenerationModel.from_pretrained("imagegeneration@002")
response = model.generate_images(
prompt="Astronaut riding a horse",
)
model.upscale_image(image=response[0])
# Upscale a new 1024x1024 image
my_image = Image.load_from_file("my-image.png")
model.upscale_image(image=my_image)
# Upscale a new arbitrary sized image using a x2 or x4 upscaling factor
my_image = Image.load_from_file("my-image.png")
model.upscale_image(image=my_image, upscale_factor="x2")
# Upscale an image and get the result in JPEG format
my_image = Image.load_from_file("my-image.png")
model.upscale_image(image=my_image, output_mime_type="image/jpeg",
output_compression_quality=90)
Args:
image (Union[GeneratedImage, Image]): Required. The generated image
to upscale.
new_size (int): The size of the biggest dimension of the upscaled
image.
Only 2048 and 4096 are currently supported. Results in a
2048x2048 or 4096x4096 image. Defaults to 2048 if not provided.
upscale_factor: The upscaling factor. Supported values are "x2" and
"x4". Defaults to None.
output_mime_type: The mime type of the output image. Supported values
are "image/png" and "image/jpeg". Defaults to "image/png".
output_compression_quality: The compression quality of the output
image
as an int (0-100). Only applicable if the output mime type is
"image/jpeg". Defaults to None.
output_gcs_uri: Google Cloud Storage uri to store the upscaled
images.
Returns:
An `Image` object.
"""
target_image_size = new_size if new_size else None
longest_dim = max(image._size[0], image._size[1])
if not new_size and not upscale_factor:
raise ValueError("Either new_size or upscale_factor must be provided.")
if not upscale_factor:
x2_factor = 2.0
x4_factor = 4.0
epsilon = 0.1
is_upscaling_x2_request = abs(new_size / longest_dim - x2_factor) < epsilon
is_upscaling_x4_request = abs(new_size / longest_dim - x4_factor) < epsilon
if not is_upscaling_x2_request and not is_upscaling_x4_request:
raise ValueError(
"Only x2 and x4 upscaling are currently supported. Requested"
f" upscaling factor: {new_size / longest_dim}"
)
else:
if upscale_factor == "x2":
target_image_size = longest_dim * 2
else:
target_image_size = longest_dim * 4
if new_size not in _SUPPORTED_UPSCALING_SIZES:
raise ValueError(
"Only the folowing square upscaling sizes are currently supported:"
f" {_SUPPORTED_UPSCALING_SIZES}."
)
instance = {"prompt": ""}
if image._gcs_uri: # pylint: disable=protected-access
instance["image"] = {
"gcsUri": image._gcs_uri # pylint: disable=protected-access
}
else:
instance["image"] = {
"bytesBase64Encoded": image._as_base64_string() # pylint: disable=protected-access
}
parameters = {
"sampleCount": 1,
"mode": "upscale",
}
if upscale_factor:
parameters["upscaleConfig"] = {"upscaleFactor": upscale_factor}
else:
parameters["sampleImageSize"] = str(new_size)
if output_gcs_uri is not None:
parameters["storageUri"] = output_gcs_uri
parameters["outputOptions"] = {"mimeType": output_mime_type}
if output_mime_type == "image/jpeg" and output_compression_quality is not None:
parameters["outputOptions"][
"compressionQuality"
] = output_compression_quality
response = self._endpoint.predict(
instances=[instance],
parameters=parameters,
)
upscaled_image = response.predictions[0]
if isinstance(image, GeneratedImage):
generation_parameters = image.generation_parameters
else:
generation_parameters = {}
generation_parameters["upscaled_image_size"] = target_image_size
encoded_bytes = upscaled_image.get("bytesBase64Encoded")
return GeneratedImage(
image_bytes=base64.b64decode(encoded_bytes) if encoded_bytes else None,
generation_parameters=generation_parameters,
gcs_uri=upscaled_image.get("gcsUri"),
)
@dataclasses.dataclass
class ImageGenerationResponse:
"""Image generation response.
Attributes:
images: The list of generated images.
"""
__module__ = "vertexai.preview.vision_models"
images: List["GeneratedImage"]
def __iter__(self) -> typing.Iterator["GeneratedImage"]:
"""Iterates through the generated images."""
yield from self.images
def __getitem__(self, idx: int) -> "GeneratedImage":
"""Gets the generated image by index."""
return self.images[idx]
_EXIF_USER_COMMENT_TAG_IDX = 0x9286
_IMAGE_GENERATION_PARAMETERS_EXIF_KEY = (
"google.cloud.vertexai.image_generation.image_generation_parameters"
)
class GeneratedImage(Image):
"""Generated image."""
__module__ = "vertexai.preview.vision_models"
def __init__(
self,
image_bytes: Optional[bytes],
generation_parameters: Dict[str, Any],
gcs_uri: Optional[str] = None,
):
"""Creates a `GeneratedImage` object.
Args:
image_bytes: Image file bytes. Image can be in PNG or JPEG format.
generation_parameters: Image generation parameter values.
gcs_uri: Image file Google Cloud Storage uri.
"""
super().__init__(image_bytes=image_bytes, gcs_uri=gcs_uri)