常见问题 | deep-imager AI 图片生成器

deep-imager is an independent, browser-based AI image generator and AI video workspace. You can run text to image, refine with image to image, pick aspect ratios and styles, then download results from the same account.

The product is built for creators who do not want to juggle separate Discord bots, multiple subscriptions, and export tools. Optional model access may include paths such as GPT-4o image, Nano Banana, Midjourney-style looks, and video models such as Sora 2 or Veo 3 where available. deep-imager is not affiliated with those trademark owners; names describe model access, style compatibility, or comparisons only.

Register or log in, open the workspace, choose text to image, write a prompt that includes subject, lighting, and framing, lock the aspect ratio for your channel, then press Generate. Credits for that run appear on the button before you confirm.

For a first successful still, keep the prompt concrete (who/what, where, camera distance, mood) and avoid stacking conflicting styles. After you like a result, use image to image to restyle or refit the canvas instead of rewriting the brief from scratch. See the user guide for a longer walkthrough.

Image to image keeps composition cues from a reference while you steer edits in natural language. Starting a brand-new text to image run invents a new layout and often wastes credits when you only needed a ratio change or style shift.

Use image to image when you already have a winner still and need pack-shot cleanup, background swap, or a 9:16 remake of a 1:1 hero. Use fresh text to image when the concept itself must change.

New accounts receive 5 free credits so you can test the AI image generator before paying. For ongoing work, buy a paid plan. Basic starts at $1/month for 10 credits; larger packs suit social batches and video tests.

Credit cost depends on model and batch size. Always read the Generate button. Failed runs do not keep the deduction — see how credits work.

No. Failed generations do not keep a lasting deduction under normal platform rules. Check status in My Works / dashboard history. If a job shows failed with an error message, credits should return according to the refund path for that failure type.

Credits are generally valid for 1 month from the date they are added. Daily check-in batches expire at the next day’s 00:00:00 boundary when that feature is active. Always confirm the expiry shown on your credit record in the dashboard.

Models and our safety layer reject prompts or outputs that hit copyrighted characters, celebrity likenesses, brand marks, or other sensitive limits. Adjust the prompt, remove protected names, and try again. A machine safety block is separate from manual gallery approval status.

Outputs are delivered without a deep-imager watermark overlay. High-resolution originals may require a membership plan. You remain responsible for not uploading or publishing infringing third-party marks inside the artwork itself.

Paid users may use eligible creations in personal and commercial projects, subject to our Terms and applicable law. Do not infringe others’ brands, likenesses, or copyrights. Review Terms of Service and plan details on Pricing before client delivery.

When you enable Publicly visible, you ask to list the still in the public gallery. Submissions go through automated safety checks and human review (pending / approved / rejected). Private generations stay in your account history and are not listed publicly.

Do not publish illegal content, sexual content involving minors, non-consensual intimate imagery, extreme gore for shock value, hate or harassment, clear copyright theft, scams, or malware. Full rules are in Terms § Public Gallery and UGC.

We combine prompt/output automated moderation with human review for public listings. We may hide, remove, or refuse to publish content and may suspend accounts that repeatedly break the rules. Machine blocks do not automatically rewrite gallery approval fields.

Use the Report this image link on a gallery detail page, or email support@flyingmob.com with the URL, image ID, and a short description. You can also start from the Contact page.

For same-day social batches, prefer a fast still model (for example Nano Banana) and lock 9:16 / 1:1 / 16:9 before generating. For product pack shots, start with a clean reference and image to image background cleanup. For mood concept art, Midjourney-style paths can help explore look — then refine with image to image for layout control.

Video teaser clips use video models where available and cost more credits than stills. Compare options on Tools and independent pages such as vs Midjourney (unaffiliated comparison).

Single-model tools force you to leave the product to change engines. deep-imager keeps text to image, image to image, styles, templates, and gallery remix (Make Same) in one browser session with a shared credit balance. That reduces context switching for teams that ship stills and short video from the same brief.

Email support@flyingmob.com or open Contact. Include your account email and any image or order IDs. For privacy and Cookie questions, see the Privacy Policy.