WEBVTT

00:00:04.000 --> 00:00:08.272
Start with an idea. Turn it into an image, then shape the details until

00:00:08.319 --> 00:00:12.185
it fits your purpose. In this chapter, we will create a campaign illustration

00:00:12.255 --> 00:00:17.061
for our fictional Willow Cafe, revise two areas, compare versions, and

00:00:17.131 --> 00:00:21.264
download the result. This demonstration uses the web app on a desktop

00:00:21.322 --> 00:00:26.059
computer.  In Studio, choose Images, then Generate. Write the brief in

00:00:26.094 --> 00:00:30.471
the Prompt field. You can start independently with an idea; you do not

00:00:30.529 --> 00:00:34.975
need a recorded conversation. In this example, the open session is not

00:00:35.022 --> 00:00:39.004
automatically included, so the prompt supplies the background.  Describe

00:00:39.027 --> 00:00:42.742
the subject, the mood, and where the picture will be used. Our brief

00:00:42.812 --> 00:00:47.793
asks for a coffee cup, a small plant, and soft morning light. We specify

00:00:47.851 --> 00:00:51.624
an editorial illustration in cream and forest green. We also ask for

00:00:51.670 --> 00:00:55.513
the objects on the right and clear space on the left, ready for a headline

00:00:55.536 --> 00:01:00.215
added later. These composition details help make the result useful.

00:01:00.215 --> 00:01:04.314
Below the prompt, check the provider and model. We are using Google Gemini

00:01:04.383 --> 00:01:09.596
with Nano Banana Two, Flash. Available options can vary. Choose an aspect

00:01:09.654 --> 00:01:14.205
ratio that fits the destination. Here, Slides, sixteen by nine, gives

00:01:14.251 --> 00:01:19.279
us a wide composition. We select two K for image size. Check the allowance

00:01:19.313 --> 00:01:22.831
shown in your account before generating.  Choose Generate and follow

00:01:22.878 --> 00:01:27.104
the progress. The app takes time to create and save the image. We have

00:01:27.150 --> 00:01:31.028
shortened the waiting in this lesson. Once the result is ready, open

00:01:31.086 --> 00:01:35.231
it to review the actual output.  Here is our first version. The coffee

00:01:35.277 --> 00:01:38.900
cup and plant sit on the right, with room for a headline on the left.

00:01:39.213 --> 00:01:43.381
The details on the right record the prompt, model and source. This one

00:01:43.416 --> 00:01:48.165
is a freeform image, with no conversation context attached. Check the

00:01:48.211 --> 00:01:52.437
composition before deciding what to change.  Give the image a recognizable

00:01:52.484 --> 00:01:57.789
name using Rename. We call this Willow Cafe, opening campaign. That makes

00:01:57.813 --> 00:02:01.644
the work easier to find when the collection grows.  Now choose Revise.

00:02:02.015 --> 00:02:05.417
Instead of starting the brief again, work from the existing picture.

00:02:05.998 --> 00:02:09.608
Think about what you want to change and what should remain. A specific

00:02:09.666 --> 00:02:13.707
instruction gives the next version a clearer direction.  Drag a rectangle

00:02:13.741 --> 00:02:17.735
over the area you want to change. We select a clear part of the tabletop.

00:02:18.107 --> 00:02:22.460
The comment asks what should change here.  Ask for a small golden croissant

00:02:22.530 --> 00:02:26.907
on a cream ceramic plate, while keeping the coffee cup and plant. Choose

00:02:26.942 --> 00:02:31.121
Done to keep this instruction with its marked area.  Mark a second area

00:02:31.179 --> 00:02:35.626
around the outdoor view. This time, ask for soft, uncluttered greenery

00:02:35.939 --> 00:02:39.875
and keep the wooden window frame. Each area gets its own instruction,

00:02:40.038 --> 00:02:44.542
so the two changes remain easy to review.  The current edits appear as

00:02:44.589 --> 00:02:49.303
a numbered list. Review them before submitting. The controls beside each

00:02:49.396 --> 00:02:53.668
entry let you edit or remove that area. Zoom controls help you inspect

00:02:53.703 --> 00:02:57.267
the picture more closely while you work.  For a change across the whole

00:02:57.302 --> 00:03:01.435
picture, use the whole-image note. We ask for slightly warmer morning

00:03:01.493 --> 00:03:05.405
light, while preserving the clear headline space and the welcoming illustration

00:03:05.452 --> 00:03:09.875
style. This is separate from the two marked areas.  Check the generation

00:03:09.945 --> 00:03:14.380
settings again. In this recording, the revision form started at a smaller

00:03:14.450 --> 00:03:19.430
image size, so we select two K again. Do not assume a new revision has

00:03:19.500 --> 00:03:24.074
kept every earlier setting.  Apply the three edits: two marked areas

00:03:24.121 --> 00:03:28.359
and one overall instruction. Let the revision finish. Again, the waiting

00:03:28.405 --> 00:03:32.875
is shortened here.  The revised image now includes the croissant, a simpler

00:03:32.933 --> 00:03:37.252
outdoor view, and warmer light. Review the entire picture, including

00:03:37.310 --> 00:03:41.594
details outside the marked regions. AI revisions can change more than

00:03:41.629 --> 00:03:45.518
the exact area you selected.  Use the version buttons to return to the

00:03:45.565 --> 00:03:49.547
original. Compare the composition and the details you wanted to preserve.

00:03:49.860 --> 00:03:53.959
Then select version two to see the revision again. Before another revision

00:03:54.028 --> 00:03:57.882
or a download, check which version is selected.  Return to the collection

00:03:57.929 --> 00:04:01.992
to find your saved work. A clear title helps you recognize it. The search

00:04:02.039 --> 00:04:05.545
field narrows the images already loaded, and the sort control changes

00:04:05.580 --> 00:04:09.493
their order. Here, we choose Title. Open the image to continue where

00:04:09.528 --> 00:04:13.417
you left off.  Download the selected version. Our actual download is

00:04:13.464 --> 00:04:17.434
a Web P image. Open the file and check the details at the size you plan

00:04:17.481 --> 00:04:22.740
to use.  From a clear brief to a finished visual: generate, mark changes,

00:04:22.903 --> 00:04:27.129
add overall direction, compare versions, and keep the result you want.

00:04:27.802 --> 00:04:30.983
Return to your image collection whenever the next idea is ready.