A photo slideshow arranges the stills you uploaded
create_slideshow_from_uploaded_images uses your photos in order. It does not generate new slide art.
A photo slideshow arranges the stills you uploaded. Your photos, in order, with captions if you want them. create_slideshow_from_uploaded_images takes the image URLs you already have. The order you give them becomes the slide order. It does not generate new art. Asking this job to "make a carousel about morning habits" without photos wastes credits on a generate-from-prompt slideshow you did not mean to buy — or it fails because there were no stills to arrange.
Turn my photos into a slideshow is the named capability. The AI slideshow maker is the studio door. Generate a slideshow from a prompt is the other slideshow: it plans and generates images. Do not mix the two because both say "slideshow."
What this job actually is
You have six vacation photos, a product set, a before-and-after pair. You upload or link them. You say the order. Optionally you ask for overlay captions and the agent writes a viral-style line per slide and stamps it. The slideshow saves with your images. There is no image-generation charge for the stills. The caption overlay is a lightweight text pass, not a full model generation.
That is why this job is cheap relative to a generated carousel, and why using it as a stand-in for a generated carousel is a category error: you either have photos or you do not.
Output is a slideshow, not a rendered reel. Converting that slideshow into a video with transitions is turn a slideshow into a video. Editing one slide later is edit an existing slideshow. Neither of those is this job.
What wastes credits
No photos, just a theme. "Make a 6-slide reel about morning habits" is create_slideshow / create_slideshow_caption_in_image. Those generate images. This tool's own description says not to use them when the user already gave photos. The reverse is also true: do not call the upload tool when you wanted generated slides.
You wanted a video in one step. Uploading stills and expecting motion, music, and a 9:16 export is two jobs. Arrange first. Render second. Skipping the slideshow object and dumping stills into a movie generate is how you pay scene prices for a carousel.
You wanted new pictures of the same trip. If the photos are wrong, replace a slide or generate a slideshow. Do not "turn these into a slideshow" and also "make them look like film stills" in one breath unless you are ready for an image-edit on each file — which is not this tool.
The test
Are the images in your hand the images you want on the slides?
If yes, this is the job. Give order. Ask for captions only if you actually want them stamped.
If you need the agent to invent the pictures, generate a slideshow from a prompt. If you already have a slideshow object and you need a reel, convert it. If you have clip URLs, you wanted an editor assembly, not slides. This is one named agent job. Using it for a different job wastes credits.
FAQ
Do my photos get regenerated as part of the slideshow?
No. The tool uses the files you provided. Order is the creative act. Optional overlay captions are text on those files, not new generations of the photos.
Can the agent write a caption per slide?
Yes, if you ask. Describe the vibe. That overlay is a text pass, not an image model. If you omit the text style, you get slides with no overlays.
Is this the same as stitching images into a video?
No. Stitch multiple clips into one video concatenates files into one video timeline. A slideshow is a slide object you can still reorder, caption, and later render. Pick the object you want to keep editing.
Why not generate nicer versions of my photos first?
You can, as a prior job, if the stills are not keepers. Then this job arranges the keepers. Generating "nicer" versions inside the slideshow request is how you lose the photos you came to arrange.