Daguerra is a studio for creators and agencies who run AI characters, on a RunPod GPU in your own account. Carousels for each post, scripts that unfold step by step, your own LoRA for the face you keep. Launch the template, paste your license key and start producing.
NewPrompt from an image in one click →
The pod runs in your own account. You choose the GPU and RunPod bills the time to you.
One studio for the posts of your characters. Carousels, scripts that move a story forward and your own LoRA, on a GPU you rent in your own RunPod account. Your images stay on your volume.
One AI character, posting across a feed, a subscription page and a few platforms. A carousel gives each post its own set of pictures in one scene and one outfit. Script mode tells a story step by step, with free and pay-per-view steps.
Creator covers the Studio and the Carousel. Pro adds your own LoRA, Script mode and the video tab.
Keep each character's LoRA on your volume and choose it for each image. Agency runs up to four pods at the same time, so several characters can be in production together.
Agency starts at $299 a month. You ask for a quote before you buy.
Launch it on RunPod with a GPU of 24 GB or more and a 120 GB volume. Daguerra runs in your account, so you pay RunPod directly for the time.
On first boot the missing nodes and models download by themselves and are checked by hash. About 50 GB, usually 30 to 60 minutes, then kept on your volume.
Write a prompt or drop a reference image, choose a format and press Generate. ComfyUI is one click away in the same header when you want the full node editor.
Four steps, and only the download takes a while. Later starts are quick, because the models stay on your volume.
Get a license →Deploy it on RunPod (new account: $5 free credit) with a GPU of 24 GB or more, and attach a 120 GB volume.
Daguerra checks the license on the pod, offline. Add your Civitai token once.
Nodes and models download on first boot and are verified by hash. The first download is about 50 GB and usually takes 30 to 60 minutes.
About 50 GB, 30 to 60 minDaguerra unlocks as soon as the pipeline is ready. Open port 3000, write a prompt and press Generate.
Live previews show the composition settling, so you can stop a weak run early and keep your GPU time for the good ones.
Next: prompts →
Features
22steps
Live preview. The default first pass, watched as it runs. Stop it early if the composition is off.
A step counter and a progress bar sit over the frame while it develops.
Press Stop and the queue is yours again, ready for the next prompt.
Switch between the finished image and the raw first pass.
Download a result, copy its prompt, or send the same settings back to the form.
Ctrl or Cmd with Enter generates. Arrow keys move between results; D downloads, R reuses the settings, C copies the prompt, F goes full screen.
Press Enhance to turn a short idea into a full description of framing, camera, light and setting. Or drop a reference image and Daguerra writes the prompt from it, then generates in one click.
Next: your own styles →
Features
Example
You writeDunes at dusk
Enhance writesVertical 3:4 view of rolling dunes at dusk, low sun on the horizon. Camera: full-frame body, 50 mm equivalent. Lighting: warm low sun, soft haze above the ridges. Finish: muted tones, light sensor grain.
Read the longer prompt, edit it, or press Undo to get your own words back.
Drop a reference image and a vision model writes the prompt. Add an optional instruction, such as a different setting, to steer it.
Generate reads the image, writes the prompt and starts at once. Turn on Review the prompt before generating to edit it first.
A phrase added to every prompt, for a look or subject you want in every image.
Upload a LoRA or pick one already on your volume. Set its strength and trigger word once, and keep the same look across a whole series of images.
Next: camera look →
Features
My LoRAs up to six
One look, kept across the whole series.
Stack several LoRAs in one image and set the strength of each.
Turn a LoRA off without removing it, and back on when you need it.
Upload a .safetensors file from the form, or choose one already in the loras folder of your volume.
Lock a seed to reproduce a result, or leave it random. Each image of a batch gets its own seed.
One button under a finished image gives it the look of a phone, a compact or a DSLR. Slide across the picture to compare before and after, and download either version.
Next: what you need →
Features
Camera look before / after
Makes the image look like it came out of a real camera: softer optics, sensor noise, light tone shifts and a JPEG finish.
Two sliders set how strong the look is and how much grain it carries.
The original is kept. Drag across the picture to compare both, and download either one.
Every result lands in a history strip under the viewer. Move between images with the arrow keys.
Besides the Studio, the same app has a Carousel tab, an Outfit tab and a Dataset tab. They are in the header of the app and use the same pod, the same settings and the same gallery.
Next: what you need →Carousel tab
Write one scene and a list of shots. Every picture shares the scene, the outfit, the place and the light, and each shot only sets its own framing. The format and your styles come from the Studio tab.
Outfit tab
Add a photo, then describe the new outfit or add a picture of it. Only the clothes are redrawn and pasted back onto your original photo, so the rest of the picture stays as it was. Drag across the result to compare.
Use only photos of yourself or of people who gave you their consent. You are responsible for the images you create; see the terms of sale.
Dataset tab
Add one anchor image and the app generates one picture for each prompt in a list of 40, all following your anchor. You can edit or switch off any prompt, set how closely the anchor is followed, and pick Fast or Maximum quality.
Use only photos of yourself or of people who gave you their consent. You are responsible for the images you create; see the terms of sale.
ComfyUI and JupyterLab run on the same pod and are linked from the Daguerra header.
The interface adapts to small screens, with ComfyUI and JupyterLab one tap away.
A News page in the app lists what changed in every release. To update, deploy a new pod from the template: your volume and models are kept.
A RunPod account, a GPU with enough memory and a volume to keep the models. Everything else is in the template.
Also
The pod runs on your account and RunPod bills the GPU time directly. No account yet? Create one through our referral link: you get $5 of credit after your first $10 top-up, at no extra cost (we get a commission).
Free, and needed once to download the main model.
The template requires it; pick a host that offers it.
Hosted image tools meter every picture. Daguerra is a flat license plus the GPU minutes you use.
| Hosted creditsNano Banana Pro on Higgsfield | Daguerraon your own RunPod | |
|---|---|---|
| Price per image | $0.08 to $0.10 | About $0.01 to $0.04 of GPU time |
| 1,000 images a month | $80 to $100 | $65 to $90 with Creator |
| 5,000 images a month | $400 to $500 | $125 to $250 |
| 10,000 images a month | $800 to $1,000 | $200 to $450 |
| Fixed cost | None | License from $49 a month |
| Same character | Reference images in each prompt | Your own LoRA, trained once and kept on your volume |
| Content filters | Platform moderation | None added by Daguerra. Adults and fictional characters only, and you stay responsible. |
| Where images live | The provider’s cloud | Your RunPod volume |
Hosted prices follow Higgsfield’s published rates checked in October 2026: Nano Banana Pro costs 2 credits an image, and credits cost about $0.04 to $0.05 each, from a $49 plan with 1,200 credits up to the 20 credits per dollar reference rate. Time-limited unlimited windows exist on higher plans, but not for automated use. Daguerra figures assume an RTX 4090 or RTX PRO 4500 at $0.72 to $0.89 an hour and 1 to 3 minutes an image depending on quality, plus the $49 Creator plan; model loading and idle time are billed too. Below about 1,000 images a month the two cost about the same. The gap opens with volume.
A license covers the Daguerra software. GPU time is billed by RunPod to your own account.
Terms of sale →The Studio and the Carousel, for a creator who posts regularly.
Every tool in the app. Or $990 a year: two months free.
Everything in Pro, for life. Limited to 20 founders.
Several pods at the same time, for teams running many characters.
Which one? A creator with one character: Creator, or Pro for your own LoRA, Script mode and the video tab. A team with several characters: Agency, with up to four pods at the same time. Founder is everything in Pro, for life, in 20 places.
Pay in USDC on Solana and your key arrives by email. Payments are checked by hand, usually within a few hours. A monthly key lasts 30 days and a yearly key 365 days. After the end date the studio keeps working 3 more days, so a late payment never stops a job; when you pay the next period we send a new key. Bought Studio for $99 before? Your key keeps working for life, and you get 50% off your first 3 months of Pro: reply to your license email.
Send us your character (a few photos or your LoRA) and what you need: carousels, a script with free and pay-per-view steps, new outfits, a product in hand. We make the images with Daguerra, check each one by hand and deliver the files without metadata.
Packs or a monthly plan, with a quote within 24 hours. Fictional characters, or adults who gave their written consent.
A Pro key valid 3 days, with every tool of the app. You only pay RunPod for the GPU time you use: a full test usually costs a few dollars.
One trial per person. The key arrives by email within a few minutes.
Written by people who use Daguerra. Every review is read before it is published.
No reviews yet. Yours can be the first, once you have used Daguerra.
Anything else is covered in the terms and refunds.
No. The pod runs in your own RunPod account, so RunPod bills you directly. A license covers the Daguerra software.
Nothing is charged automatically. In the last week the app shows a bar asking you to renew, and after the end date it keeps working for 3 more days. Pay the next period on the purchase page, paste the new key under License, and you carry on. If you do not renew, the studio stops generating; your images and models stay on your volume.
You receive a key by email after purchase and paste it into Daguerra, or set it as LICENSE_KEY on the template. The pod checks it with our license server at start-up and about once an hour; one key runs one pod at a time.
From their original sources, Civitai and Hugging Face, with your own tokens. Each model keeps its own license, which you accept with your own account.
Yes. ComfyUI and JupyterLab run on the same pod and are linked from the Daguerra header.
Yes, on the Founder and Agency plans. Your pod has an MCP address (the app address plus /mcp) that you protect with the API token from the License screen. An assistant can write a prompt, start images, carousels and scripts, then fetch the results. Outfit, Dataset and Animate stay in the app.
They are generated and stored on your pod, in your volume. Daguerra does not upload them anywhere.
The template does not depend on a datacenter. Pick another GPU or region on RunPod and attach a volume in that region.
Inside one script, yes: each step is drawn from the previous picture, so the person and the place stay the same. Across separate sets the look can drift, so keep your own LoRA for the character (Pro, Founder and Agency). The guide is strong, not a guarantee: check every picture before you post it.
Yes. Keep each character's LoRA on your volume and choose the ones you need in My LoRAs, up to six per image. Pro, Founder and Agency include My LoRAs. Agency runs up to four pods at the same time.
No. Daguerra makes the pictures on your pod and keeps them on your volume. You download and publish them yourself, and you follow the rules of each platform, including its rules on AI-generated images. Each model keeps its own license, and some limit commercial use, so read them before you sell an image.
Fictional characters, or adults who have agreed to it. You must be 18 or older. Never a real person without their consent, and never sexual or deceptive images of real people. The terms of sale have the full rules.
Get a license key and deploy the template on your RunPod account.
