A document may span many files
Bundled DjVu stores its page data together. An indirect document has a small index and separate page files. Uploading only the index cannot reproduce the whole book.
Bundle before conversion
Use a DjVu-capable desktop tool to bundle an indirect document while all its referenced files are available. This reader deliberately avoids following external references.
Use a focused page range
A book can contain thousands of pages while a browser has limited working memory. Select a useful chapter or a small range, then continue with the next range after reviewing the output.
Work through a small chapter first
Start with a copy of the book and select pages 12-16. Run the converter, confirm that the result contains five pages, then compare the first preview with source page 12 in your existing DjVu reader. Page numbers printed inside the scan may differ from the document page positions used by the range field.
For a 240-page book, an empty range is rejected because the workbench permits at most 50 selected pages. Enter an explicit chapter range rather than assuming the first 50 pages will be chosen. The source can contain up to 2,000 pages, but the selected image dimensions also have to fit the pixel limits.
A very small index file next to many similarly named page files is a clue that the document is indirect. Do not rename that index or zip the directory and expect it to become bundled DjVu. Use a DjVu-capable application to create a complete bundled document, then reopen that new file before uploading it here.
If an unsupported-codec message appears, keep the exact codec name with your troubleshooting notes. It means this decoder cannot reproduce that source, not that the whole book is corrupt. A complete desktop DjVu implementation may open it successfully; this workbench deliberately stops instead of returning a partially decoded book.