Creating text to image for panoramic views

Hi, I tried several models but did not succeed in creating a 1500 x 200 cm - 72 dpi panoramic view.
Because of the limited dimensions models can create, I tried generate 9 panels that could be seamed together afterward in Photoshop. I copy the prompt I used for this her. Anyone who could be able to solve this ?
Panel 1 - left 1

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light shining on a dramatic mountain range,

snow-covered pine trees in the foreground,

a cozy Christmas mountain village glowing with warm lights in the distance,

cinematic, highly detailed, 720-dpi look,

panel 1 of 9 – far left of a continuous panorama,

extra scenery extending off the left edge,

ample overlap on the right to connect seamlessly with panel 2,

consistent color grading and lighting.

Panel 2 - left 2

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light on a dramatic mountain range,

snow-covered pine trees, hints of the alpine village lights,

cinematic, highly detailed, 720-dpi look,

panel 2 of 9 – left side of a continuous panorama,

elements matching panel 1 on the left and blending toward panel 3 on the right,

consistent color grading and lighting.

Panel 3 - left 3

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light on mountain peaks,

dense snow-covered forest and parts of the glowing Christmas village,

cinematic, highly detailed, 720-dpi look,

panel 3 of 9 – center-left of a continuous panorama,

smooth transition from panel 2 on the left and toward panel 4 on the right,

consistent color grading and lighting.

Panel 4 - central 1

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light on mountain peaks,

snowy evergreens and cozy village lights,

cinematic, highly detailed, 720-dpi look,

panel 4 of 9 – just left of center in a continuous panorama,

perfect overlap with panels 3 and 5,

consistent color grading and lighting.

Panel 5 – central 2

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light striking the tallest mountains,

the heart of the Christmas village with warmly lit chalets and smoke from chimneys,

cinematic, highly detailed, 720-dpi look,

panel 5 of 9 – center of a continuous panorama,

balanced composition connecting smoothly to panels 4 and 6,

consistent color grading and lighting.

Panel 6 - central 3

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light on mountain peaks,

snow-covered village rooftops and pine trees,

cinematic, highly detailed, 720-dpi look,

panel 6 of 9 – center-right of a continuous panorama,

seamless continuation from panel 5 on the left toward panel 7 on the right,

consistent color grading and lighting.

Panel 7 - right 1

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light, snowy forest and scattered village lights,

cinematic, highly detailed, 720-dpi look,

panel 7 of 9 – right-center of a continuous panorama,

overlap with panels 6 and 8 for perfect alignment,

consistent color grading and lighting.

Panel 8 - right 2

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light fading into cool blue twilight,

more distant mountains and snow-covered trees,

cinematic, highly detailed, 720-dpi look,

panel 8 of 9 – right side of a continuous panorama,

matching panel 7 on the left and blending toward panel 9 on the right,

consistent color grading and lighting.

Panel 9 – right 3

Ultra-wide panoramic photograph of an alpine snow landscape at late evening,

golden hour light fading into twilight,

snowy trees and open mountain view extending off the right edge,

cinematic, highly detailed, 720-dpi look,

panel 9 of 9 – far right of a continuous panorama,

ample overlap on the left to connect seamlessly with panel 8,

consistent color grading and lighting.

1 Like

At 72dpi, the resulting image would be enormous in both height and width, requiring massive amounts of VRAM to generate it directly with AI.

Therefore, a more practical approach would be to first generate a blurry, horizontal overview image. Then, using that as a reference, apply upscaling with a Tile ControlNet to refine and clean up the details.