tcsenpai/ollama
1
0
Fork 0
mirror of https://github.com/tcsenpai/ollama.git synced 2025-06-07 11:45:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

add compositing for pngs

Browse Source
This commit is contained in:
Patrick Devine 2024-10-08 18:46:58 -07:00
parent 3a1c8da5e4
commit 6cb0abf6d8
2 changed files with 20 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
server/imageproc/images.go
View File

@ -4,6 +4,7 @@ import (
"bytes" "bytes"
"fmt" "fmt"
"image" "image"
"image/color"
_ "image/jpeg" _ "image/jpeg"
_ "image/png" _ "image/png"
"math" "math"
@ -139,7 +140,22 @@ func splitToTiles(img image.Image, numTilesSize image.Point) []image.Image {
return images return images
} }
func ResizeImage(img image.Image, outputSize image.Point, maxImageTiles int) (image.Image, image.Point) { // remove the "alpha" channel by drawing over a prefilled image
func compositeImage(img image.Image) image.Image {
dst := image.NewRGBA(img.Bounds())
white := color.RGBA{255, 255, 255, 255}
draw.Draw(dst, dst.Bounds(), &image.Uniform{white}, image.Point{}, draw.Src)
draw.Draw(dst, dst.Bounds(), img, img.Bounds().Min, draw.Over)
return dst
}
func ResizeImage(img image.Image, format string, outputSize image.Point, maxImageTiles int) (image.Image, image.Point) {
if format == "png" {
img = compositeImage(img)
}
b := img.Bounds() b := img.Bounds()
tileSize := outputSize.Y tileSize := outputSize.Y
@ -217,12 +233,12 @@ func Preprocess(imageData []byte) ([]float32, int, error) {
mean := [3]float32{0.48145466, 0.4578275, 0.40821073} mean := [3]float32{0.48145466, 0.4578275, 0.40821073}
std := [3]float32{0.26862954, 0.26130258, 0.27577711} std := [3]float32{0.26862954, 0.26130258, 0.27577711}
img, _, err := image.Decode(bytes.NewReader(imageData)) img, format, err := image.Decode(bytes.NewReader(imageData))
if err != nil { if err != nil {
return nil, 0, fmt.Errorf("failed to decode image: %w", err) return nil, 0, fmt.Errorf("failed to decode image: %w", err)
} }
newImage, aspectRatio := ResizeImage(img, outputSize, maxTiles) newImage, aspectRatio := ResizeImage(img, format, outputSize, maxTiles)
newImage = PadImage(newImage, outputSize, aspectRatio) newImage = PadImage(newImage, outputSize, aspectRatio)
data := PackImages(newImage, aspectRatio, mean, std) data := PackImages(newImage, aspectRatio, mean, std)

2
server/imageproc/images_test.go
View File

@ -246,7 +246,7 @@ func TestResize(t *testing.T) {
} }
for _, c := range cases { for _, c := range cases {
actualImage, actualAspectRatio := ResizeImage(c.TestImage, c.OutputSize, c.MaxImageTiles) actualImage, actualAspectRatio := ResizeImage(c.TestImage, "png", c.OutputSize, c.MaxImageTiles)
if actualImage.Bounds() != c.ExpectedImage.Bounds() { if actualImage.Bounds() != c.ExpectedImage.Bounds() {
t.Errorf("image size incorrect: '%#v': expected: '%#v'", actualImage.Bounds(), c.ExpectedImage.Bounds()) t.Errorf("image size incorrect: '%#v': expected: '%#v'", actualImage.Bounds(), c.ExpectedImage.Bounds())