Style-based dynamic content generation
Systems, devices, and techniques are disclosed for style-based dynamic content generation. A seed image, entity design data, entity style data, and text items may be received. Bounding boxes that identify areas of the seed image for the placement of the text items may be generated for the seed image. Variant images may be generated from the seed image, the entity design data, and the entity style data. The variant images may be generated by placing text items in the bounding boxes based on the entity design data and rendering text of the text items using the entity style data.
1 . A computer-implemented method comprising:
receiving a seed image, entity design data, entity style data, and one or more text items;
generating for the seed image one or more bounding boxes that identify one or more areas of the seed image for the placement of the one or more text items;
generating from the seed image, the entity design data, and the entity style data two or more variant images, wherein each of the variant images is generated by placing text items of the one or more text items and one or more items of iconography in one or more of the one or more bounding boxes, wherein a style of at least one of the one more or text items is based on a position of the at least one of the one or more text items and entity styles for text in different positions from the entity design data, positioning of at least one of the one or more text items relative to at least one of the one or more items of iconography is based on the entity design data and wherein text of the text items is rendered using the entity style data, and further comprising adjusting an image spectra of at least one of the one or more bounding boxes based on the image spectra of the one of the one more bounding boxes and a color of one of the one or more text items in the one of the one or more bounding boxes;
generating at least one additional bounding box for the seed image to cover an area between a boundary of one of the one or more bounding boxes and a horizontal or vertical edge of the seed image.
2 . The computer-implemented method of claim 1 , wherein the entity style data comprises one or more of font faces, font sizes, font style, font weight, colors, and spacing.
3 . The computer-implemented method of claim 1 , further comprising generating the one or more text items using a text style transformer model trained using text items from the entity.
4 . The computer-implemented method of claim 1 , wherein generating for the seed image one or more bounding boxes that identify one or more areas of the seed image for the placement of the one or more text items comprises inputting the seed image to a one of object detection model or a generative adversarial network.
5 . The computer-implemented method of claim 1 , further comprising:
generating, from the seed image, a seed image variant;
generating for the seed image variant one or more bounding boxes that identify one or more areas of the seed image variant for the placement of the one or more text items; and
generating from the seed image variant, the entity design data, and the entity style data two or more additional variant images, wherein each of the additional variant images is generated by placing text items of the one or more text items in one or more of the one or more bounding boxes, generated for the seed image variant, based on the entity design data and wherein text of the text items is rendered using the entity style data.
6 . The computer-implemented method of claim 5 , wherein generating the seed image variant comprises cropping the seed image such that an object of the seed image is not cropped out.
7 . The computer-implemented method of claim 1 , further comprising generating additional sets of bounding boxes for the seed image by adjusting at least one edge of at least one of the one or more bounding boxes.
8 . The computer-implemented method of claim 1 , wherein generating from the seed image, the entity design data, and the entity style data two or more variant images, comprises using the seed image, bounding boxes, entity style data, and one or more text items as input to a generative adversarial network.
9 . The computer-implemented method of claim 1 , wherein the entity design data comprises statistical data gathered from content associated with an entity from which the entity style data was gathered.
10 . A computer-implemented system comprising:
one or more storage devices; and
a processor that receives a seed image, entity design data, entity style data, and one or more text items,
generates for the seed image one or more bounding boxes that identify one or more areas of the seed image for the placement of the one or more text items,
generates from the seed image, the entity design data, and the entity style data two or more variant images, wherein each of the variant images is generated by placing text items of the one or more text items and one or more items of iconography in one or more of the one or more bounding boxes, wherein a style of at least one of the one more or text items is based on a position of the at least one of the one or more text items and entity styles for text in different positions from the entity design data, positioning of at least one of the one or more text items relative to at least one of the one or more items of iconography is based on the entity design data and wherein text of the text items is rendered using the entity style data, and further adjusts an image spectra of at least one of the one or more bounding boxes based on the image spectra of the one of the one more bounding boxes and a color of one of the one or more text items in the one of the one or more bounding boxes, and
generates at least one additional bounding box for the seed image to cover an area between a boundary of one of the one or more bounding boxes and a horizontal or vertical edge of the seed image.
11 . The computer-implemented system of claim 10 , wherein the entity style data comprises one or more of font faces, font sizes, font style, font weight, colors, and spacing.
12 . The computer-implemented system of claim 10 , wherein the processor further generates the one or more text items using a text style transformer model trained using text items from the entity.
13 . The computer-implemented system of claim 10 , wherein the processor generates for the seed image one or more bounding boxes that identify one or more areas of the seed image for the placement of the one or more text items by inputting the seed image to a one of object detection model or a generative adversarial network.
14 . The computer-implemented system of claim 10 , where the processor further generates, from the seed image, a seed image variant,
generates for the seed image variant one or more bounding boxes that identify one or more areas of the seed image variant for the placement of the one or more text items, and
generates from the seed image variant, the entity design data, and the entity style data two or more additional variant images, wherein each of the additional variant images is generated by placing text items of the one or more text items in one or more of the one or more bounding boxes, generated for the seed image variant, based on the entity design data and wherein text of the text items is rendered using the entity style data.
15 . The computer-implemented system of claim 14 , wherein the processor generates the seed image variant by cropping the seed image such that an object of the seed image is not cropped out.
16 . The computer-implemented system of claim 10 , wherein the processor further generates additional sets of bounding boxes for the seed image by adjusting at least one edge of at least one of the one or more bounding boxes.
17 . The computer-implemented system of claim 10 , wherein the processor generates from the seed image, the entity design data, and the entity style data two or more variant images, by using the seed image, bounding boxes, entity style data, and one or more text items as input to a generative adversarial network.
18 . The computer-implemented system of claim 10 , wherein the entity design data comprises statistical data gathered from content associated with an entity from which the entity style data was gathered.
19 . A system comprising: one or more computers and one or more storage devices storing instructions which are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
receiving a seed image, entity design data, entity style data, and one or more text items;
generating for the seed image one or more bounding boxes that identify one or more areas of the seed image for the placement of the one or more text items;
generating from the seed image, the entity design data, and the entity style data two or more variant images, wherein each of the variant images is generated by placing text items of the one or more text items and one or more items of iconography in one or more of the one or more bounding boxes, wherein a style of at least one of the one more or text items is based on a position of the at least one of the one or more text items and entity styles for text in different positions from the entity design data, positioning of at least one of the one or more text items relative to at least one of the one or more items of iconography is based on the entity design data and wherein text of the text items is rendered using the entity style data, and further comprising adjusting an image spectra of at least one of the one or more bounding boxes based on the image spectra of the one of the one more bounding boxes and a color of one of the one or more text items in the one of the one or more bounding boxes; and
generating at least one additional bounding box for the seed image to cover an area between a boundary of one of the one or more bounding boxes and a horizontal or vertical edge of the seed image.
20 . The system of claim 19 , wherein the instructions further comprise instructions that cause the one or more computers to perform operations comprising:
generating, from the seed image, a seed image variant;
generating for the seed image variant one or more bounding boxes that identify one or more areas of the seed image variant for the placement of the one or more text items; and
generating from the seed image variant, the entity design data, and the entity style data two or more additional variant images, wherein each of the additional variant images is generated by placing text items of the one or more text items in one or more of the one or more bounding boxes, generated for the seed image variant, based on the entity design data and wherein text of the text items is rendered using the entity style data.